232 results about "Semantic difference" patented technology

The invention discloses a multi-label electrocardiogram classification method based on self-supervised pre-training and multi-modal semantic alignment, which belongs to the technical field of artificial intelligence, and comprises the following steps: realizing self-supervised pre-training of unlabeled data through a single-modal contrast enhancement network, generating global and local contrast views by adopting a multi-scale random cutting strategy, and classifying the global and local contrast views in a multi-scale random cutting mode; in combination with a teacher-student network architecture, the potential invariance features of the ECG signals are learned while negative sample dependence is avoided, the problem of annotation data scarcity is effectively relieved, and the feature robustness is improved. A multi-modal fusion mechanism based on label semantic guidance is provided, a time domain signal and a frequency domain time-frequency graph are mapped to a unified semantic space through fine-grained semantic alignment, local feature enhancement and cross-modal complementary information fusion are realized by using a cross attention mechanism, and the problem of semantic difference caused by modal heterogeneity in a traditional method is overcome. A multi-label comparison loss function based on a disease co-occurrence relation is proposed, a category discrimination boundary is dynamically optimized by modeling a label co-occurrence probability, the feature separability of a tail category is improved while the head category discrimination ability is enhanced, and the problem of sample category imbalance in a multi-label scene is remarkably relieved.

The invention provides a lightweight real-time semantic segmentation method for a three-dimensional Gaussian scene, and belongs to the field of computer vision and three-dimensional scene understanding. The process comprises the following steps: S1, data preparation: acquiring multi-modal information data including point cloud, images and poses, and preprocessing the data; s2, carrying out two-dimensional semantic segmentation, and generating a pixel-level semantic segmentation map and a semantic tag for the image by utilizing PP-LiteSeg; s3, performing three-dimensional scene modeling and semantic projection, fusing multi-modal information to complete three-dimensional Gaussian scene reconstruction, and projecting two-dimensional semantics to a Gaussian function to form preliminary Gaussian function semantic features; and S4, semantic optimization and segmentation output: considering image difference, semantic difference and smoothness, establishing a loss function, optimizing semantic features and outputting a segmentation result. According to the method, efficient, real-time and accurate three-dimensional Gaussian semantic segmentation work can be completed, and the method is suitable for scene understanding tasks of light-weight and resource-limited equipment.

The invention provides a three-dimensional data synthesis method, electronic equipment, a storage medium and a program product, and the method comprises the steps: obtaining a plurality of texture-free three-dimensional grid models, carrying out the rendering of each three-dimensional grid model, obtaining a plurality of view images, and constructing a training sample pair in combination with a semantic description text; based on the training sample pair, utilizing a low-rank adaptation technology to carry out fine tuning on the first text graph model to obtain a second text graph model with multi-view consistency understanding ability; modeling a latent space semantic difference between the rendered image of the to-be-deformed three-dimensional grid model and the target semantic text based on the second text graph model, constructing an optimization constraint and updating parameters, and obtaining a deformed three-dimensional grid model conforming to target semantics; and generating a texture image by using the texture generation model, and attaching the texture image to the surface of the deformed three-dimensional grid model to obtain synthetic three-dimensional data with textures. The reconstruction precision, the diversity expression ability and the cross-category generalization ability of the synthesized three-dimensional data are obviously enhanced.

The invention belongs to an image processing technology, and particularly relates to a deep forgery detection method and system based on face embedded difference guidance, and the method comprises the steps: generating a forgery image according to an input real image, and calculating a difference image and a semantic difference vector of two images; learning the features of the real image and the forged image by using a first classifier, and learning the features of the real image and the difference image by using a second classifier; splicing and fusing the semantic difference vector and image features extracted from the difference image through a channel to obtain fused features, and learning the difference between the real face features and the fused features by using a third classifier; and calculating the distillation loss between the fusion feature and the image feature of the forged image, and training in combination with the loss of each classifier. According to the method, multi-level feature fusion, a three-classifier architecture and a knowledge distillation mechanism are adopted, the strong generalization ability across data sets is achieved, and the system can output a high-precision deep forgery judgment result.

The invention discloses a multi-scene application data acquisition method based on atomization design. The method comprises the following steps: analyzing reported data into an atomization data structure; performing semantic recognition on the data labels to generate a label semantic mapping result; constructing a layered cache structure; generating a compressed semantic mapping rule by applying a multi-level semantic difference coding algorithm; identifying an optimal query path from the tag affinity matrix; and executing historical data cleaning based on the data value. Intelligent mapping of data labels is achieved through semantic recognition and version control, the storage efficiency is improved through predictive lazy loading and compression coding, the query path is optimized based on the affinity matrix, and the technical problem of multi-scene application data collection is effectively solved.

The invention relates to a lightweight AI outbound model hot switching framework construction method. Defining a runtime context state unit to form an independent data structure; the AI model is divided into modules responsible for process reasoning execution; the module is in communication with the module responsible for dialogue memory and intention tracking state scheduling through a lightweight intermediate interface; establishing a set of semantic alignment mapping table based on the semantic difference between the models; performing semantic label conversion on a current dialogue state through a mapping table, and injecting a pseudo state vector into the new model to reconstruct an initial reasoning condition; calculating a tension score based on the intention hopping frequency, the slot position switching frequency and the user reply uncertainty; if the controller judges that the switching window is a safe switching window, hot switching logic can be triggered; a prediction module is introduced before switching to perform rehearsal on the reachable intention coverage of the new model, and switching is delayed if the risk is too high. The whole model hot switching process is not perceived by users, dialogue continuity and service continuity are guaranteed, the system resource utilization rate is improved, and model scheduling intelligence is enhanced.

The invention discloses a retrieval enhancement question and answer generation method and system based on a large language model in combination with a multi-agent cooperation mechanism. The method comprises the steps of S1, text preprocessing and semantic difference enhancement and amplification; s2, checking and complementing user questions; s3, evaluating problem complexity and selecting a generator; s4, text retrieval and answer generation; the system comprises an information enhancement processing module, a query understanding optimization module, a question complexity intelligent evaluation module and a collaborative answer generation module, and is used for realizing the method. According to the method, a multi-agent collaborative RAG framework of an information enhancement agent, an interaction analysis agent and a problem complexity evaluation agent is introduced, and semantic difference enhancement and amplification, query integrity verification and complementation and a dynamic generation strategy based on problem complexity are performed after hierarchical document partitioning are combined; and the performance of the system in the aspects of semantic distinguishing capability, query understanding precision, generation efficiency and reliability is comprehensively improved.

The invention discloses a multi-modal heterogeneous medical equipment data fusion and decision support method and device, and aims to solve the problems that the fusion precision is low due to space-time semantic difference of medical equipment multi-modal heterogeneous data (equipment operation parameters, clinical records, fault signals and the like), equipment management decisions depend on experience, and standards are not uniform. According to the method, breakthrough is achieved through three-level data alignment of'time-space-semantics', hierarchical fusion of'data level-feature level-decision level ', three-level decision driven by a knowledge graph and dynamic feedback optimization: time alignment uses a dynamic time warping algorithm, space alignment depends on a unified data dictionary, and semantic alignment introduces an attention mechanism; the feature level fusion quantifies the feature support degree based on the D-S evidence theory; the decision-making layer constructs a'rule-case-prediction 'three-level system, and combines cosine similarity retrieval and information entropy quantification uncertainty. The method and device can support medical equipment maintenance, clinical diagnosis and treatment and other scenes, and the medical service standardization level and the equipment management efficiency are improved.

The invention discloses a large model training data enhancement method, which comprises the following steps of: firstly, on the basis of an original training sample, randomly selecting a sample, generating a new training sample by using a large model, introducing a structured coding mechanism on the basis, and converting the original and generated samples into structured coding vectors; modeling and quantifying the semantic relationship between the generated sample and the original sample to obtain a query response representation reflecting the degree of semantic difference between the generated sample and the original sample; furthermore, a semantic drift degree estimated value is obtained through feature decoding, and whether the generated data is qualified or not is automatically judged based on comparison between the estimated value and a preset threshold value. Through the mode, accurate evaluation of semantic consistency between new and old training data is realized, and the problem that new data deviates from a task target due to semantic drift is effectively avoided, so that a high-quality enhanced sample which is really helpful for improving the generalization ability and robustness of a model is screened out.

The invention discloses a three-dimensional medical image registration method based on semantic difference modeling and a double-branch Transform, and relates to the field of medical image processing. According to the method, local structure keeping, global semantic modeling and boundary semantic guiding mechanisms are combined, the problem that the receptive field of a traditional CNN structure is insufficient is solved, the defect of Transform in the aspect of boundary modeling is overcome, and the method is a high-performance three-dimensional medical image registration scheme suitable for various clinical complex registration scenes; according to the network disclosed by the invention, a double-flow feature extraction mechanism and a Transform structure are introduced, so that local structure modeling and global semantic perception are effectively considered; meanwhile, the semantic diffusion fusion strategy enables the decoding process to be more accurate and robust, the decoding process is particularly superior in the aspects of complex organ deformation and boundary alignment, and a powerful end-to-end solution is provided for a three-dimensional medical image registration task.

The invention relates to the technical field of artificial intelligence, and provides a training method and device for a voice recognition large model, a medium and equipment, and the method comprises the steps: obtaining a training data set; inputting the training data set into the initial large model, and performing identification processing on the audio sample through a streaming identification branch in the initial large model to obtain a first candidate text set; according to the text consistency between the target candidate text in the first candidate text set and the real labeled text, determining whether to activate a non-streaming recognition branch to generate a second candidate text set; and performing iterative training on the initial large model according to the semantic difference between each second candidate text and the real labeled text and the text length difference between each second candidate text and the target candidate text to obtain a trained united streaming and non-streaming speech recognition large model. According to the method and the device, the problem of subtitle jumping caused by secondary identification length change can be relieved while the identification accuracy is improved.

The invention relates to an adversarial collaborative retrieval enhancement method, provides an adversarial collaborative retrieval enhancement framework (AC-RAG), and improves the performance of a large language model (LLMs) in a natural language processing (NLP) task in a specific field. In the pre-detection stage, a detector preliminarily judges whether a task needs to be retrieved or not; in the question analysis stage, the detector decomposes query into sub-tasks and provides retrieval clues for the answers; in the retrieval and integration stage, the answers retrieve and integrate information according to the subtasks; in the post-detection stage, the detector evaluates the information integrity, and retrieval is repeated if necessary. According to the framework, a neutral regulator is introduced, cooperation among agents is optimized, the problems of retrieval illusion and semantic difference are remarkably reduced, and retrieval accuracy and system efficiency are improved. In addition, in the model optimization stage, a detector and an answering device are finely adjusted through reasoning path data, so that the model is evolved continuously. Experimental results show that the AC-RAG disclosed by the invention exceeds the existing method in a plurality of vertical fields, and shows excellent performance of the AC-RAG in complex query.

The invention discloses an integrated power box health state assessment method based on AI modeling. The method comprises the following steps: S1, constructing a multivariable time sequence; s2, modeling the multivariable time sequence trajectory by adopting a neural control differential equation model to obtain a hidden space trajectory; s3, performing dynamic modal decomposition on the hidden space trajectory, and constructing a modal health space; s4, adopting a manifold interpolation mixing method in the modal health space to generate an enhanced training sample set; s5, inputting the mixed mode vector in the enhanced training sample set into a health classification module; s6, inputting the hidden space trajectory into a time semantic playback module to generate a final prediction state trajectory; and S7, comparing the final predicted state trajectory with the actual historical state trajectory, and when the semantic difference value exceeds a set tolerance threshold value, generating an abnormal mark and executing a state rollback operation. According to the method, neural differential modeling, interpolation enhancement and time semantic playback are fused, and the health state of the integrated power box is accurately evaluated.

The embodiment of the invention provides a management method and system for a design rule base of a photovoltaic power station, and the method comprises the steps: obtaining a rule document and rule data through a heterogeneous data source collection interface to form a rule base, and carrying out the repeatability verification of the collected rule base based on a file hash value and a timestamp; semantic analysis is carried out on the verified rule base, key terms, constraint conditions, an application range and a logic relation in the rule base are extracted, a rule knowledge graph is constructed, nodes represent rule terms, and edges represent the incidence relation of the rule terms; when it is detected that the rule base is updated, the semantic difference degree is calculated by comparing text differences, a new rule base version number is generated through progressive increase, change content is recorded, a change log is generated, and a rule change notification is sent to a user. According to the method, the problems of rule updating lag and inconvenient management are solved, the long-term pain point in industry rule management is solved, and the method has wide application prospect and market value.

The invention discloses an instruction checking method and device based on semantic matching, terminal equipment and a storage medium, and belongs to the technical field of instruction checking. The method comprises the steps that a current scheduling instruction and an original scheduling instruction are input into an instruction checking model, according to the method, the instruction checking model is used for respectively capturing semantic information of the current scheduling instruction and the original scheduling instruction, so that a plurality of different semantic embedding vectors are obtained, the semantic deviation degree of the current instruction and the original instruction is obtained by comparing the plurality of groups of semantic embedding vectors of the current instruction and the original instruction, and different checking prompt information is generated according to different deviation conditions; and a dispatcher is prompted to correct the instruction with semantic deviation in time. Therefore, the semantic level of the instruction can be deeply understood, the semantic difference between the instructions can be accurately recognized, the accuracy of instruction checking is improved, and the problem that the accuracy of instruction checking is low due to the fact that the semantics of the instruction cannot be deeply understood in the prior art can be solved.

Provided are a model distillation method and apparatus. The method comprises: acquiring a fixed training dataset, wherein the fixed training dataset comprises a plurality of training text sequences; guiding, by means of a teacher model, a student model to generate target response text sequences on the basis of the training text sequences, so as to correct semantic differences between response text sequences output by the student model and response text sequences output by the teacher model; on the basis of the plurality of training text sequences and a target response text sequence corresponding to each of the plurality of training text sequences, obtaining a target training dataset; and on the basis of the target training dataset and the teacher model, performing distillation training on the student model. In the present application, semantic differences between outputs of the teacher model and outputs of the student model are marked and evaluated one by one at a sequence level, and errors are identified and corrected, such that the student model is guided to generate more reliable and diversified samples, thereby improving the training effect of the student model.

The invention belongs to the technical field of industrial Internet of Things data processing, and particularly discloses a data protocol conversion method and system based on the industrial Internet of Things, and the method comprises the steps: extracting a feature vector through a protocol feature recognition model according to an original data stream of target equipment, and obtaining a matching result of a dynamic protocol fingerprint database; according to a matching result of the dynamic protocol fingerprint database, judging whether a protocol learning mode is triggered to obtain a newly added fingerprint and an analysis rule; according to the analyzed structured source data and the domain ontology library, semantic annotation is carried out through a semantic understanding engine, and semantic data with physical quantity tags are obtained; and according to the real-time resource state parameters, a conversion strategy is dynamically selected, an unloading point is decided and calculated, and a lightweight and full-amount depth conversion instruction is obtained. The invention aims to solve the problem of low protocol conversion efficiency caused by protocol isomerism, semantic difference and resource constraint in the industrial Internet of Things in the prior art.

The invention provides a translation difficulty assessment method and device based on a back-translated text, equipment and a medium, and belongs to the technical field of text translation.The translation difficulty assessment method comprises the steps that an original text is obtained, the original text is translated into a target language from a source language through a translation channel, and the translated text is obtained; the method comprises the steps of translating a translated text from a target language to a source language through a translation channel to obtain a back-translated text, calculating semantic similarity between an original text and the back-translated text, determining a translation mode of the translation channel, and determining a translation difficulty score from the original text to the target language according to a semantic difference coefficient, the semantic similarity and the translation mode. The efficiency of text translation difficulty assessment can be improved.

The invention belongs to the field of artificial intelligence and natural language processing, and discloses a cross-topic rumor detection method based on a time perception attention mechanism, and the method comprises the steps: constructing an input sample containing a text, a timestamp and a topic label based on multi-source social media data, and carrying out the semantic coding of the text through a pre-training language model; a topic attention enhancement mechanism is introduced, semantic differences of samples in different topics are captured, and the perception ability of a topic structure is improved; a time perception attention mechanism is introduced, and the robustness of time sequence characteristics is enhanced by modeling a sample time interval and dynamically adjusting an information weight; dividing positive and negative sample pairs by using topic tags, and guiding the model to learn topic-independent discriminative representation by comparing a loss function; and performing rumor judgment on the features fused with the multi-dimensional information. The method has good cross-topic migration ability and time-sensitive modeling ability, and can significantly improve the detection performance in the early propagation stage of unknown topics.

The invention provides a confidence evaluation method and device, electronic equipment and a storage medium. The method comprises the steps that key information and semantic features of a query instruction are accurately captured by query features; the search result features comprehensively represent the core key points and the text features of the document; semantic difference features are determined based on the semantic difference between the query instruction and the search result, text difference features are determined based on the text difference between the query instruction and the search result, and the internal relation between the query instruction and the search result in the semantic and text level is deeply mined; the query features, the search result features, the semantic difference features and the text difference features can comprehensively reflect the matching degree and quality information of the query instruction and the search result, so that confidence coefficient prediction is performed based on the query features, the search result features, the semantic difference features and the text difference features to obtain a confidence coefficient score result; the performance of the RAG retrieval module can be efficiently evaluated, and dependence on external reference data and a complex large language model is got rid of.

The invention discloses an Internet of Things real-time task distribution system based on AI scheduling, and the system comprises a data collection module which is used for generating a task input vector, and constructing a node resource vector set; the semantic modeling module is used for inputting the task input vector into an improved StructBERT model to construct a task semantic vector; the intention analysis module is used for constructing a historical task intention vector library and a semantic conflict graph; the comparison judgment module is used for executing task conflict judgment and scheduling priority adjustment operation according to the task aggregation degree and the semantic difference vector; and the scheduling execution module is used for executing a scheduling calculation operation, generating a task allocation result and updating the historical task intention vector library and the semantic conflict graph. The method has the advantages of high task semantic understanding precision, high conflict judgment efficiency and high scheduling decision intelligent degree, and is suitable for a real-time task allocation scene in a multi-node heterogeneous Internet of Things environment.

The invention relates to the technical field of ancient book processing and large models, and discloses a traditional Chinese medicine ancient book marking method for model adaptive optimization by fusing multi-dimensional context, which comprises the following steps: extracting multi-dimensional context features corresponding to traditional Chinese medicine ancient book texts; according to the multi-dimensional context features, dynamically determining a low-rank self-adaptive rank of at least one level in the target language model through a hierarchical rank self-adaptive adjustment strategy, and performing fine adjustment to obtain an optimized target language model; and inputting the traditional Chinese medicine ancient book text and the multi-dimensional context features into the optimized target language model, generating a preliminary labeling result, performing correction by applying a rule engine, and outputting a final labeling result. According to the method, the connotation of the ancient book text can be understood more deeply, the semantic difference of the same term in different genres or different dynasties is accurately identified, and the labeling precision of cross-genre and cross-age ancient books is improved.

The invention discloses a semantic unification and anomaly traceability analysis method and system for heterogeneous program logs, and the method comprises the steps: firstly collecting original logs in various formats, and packaging the original logs into standard objects in a unified manner; thirdly, identifying fields through a rule and a semantic model, uniformly mapping and labeling semantic tags, eliminating semantic differences, and constructing log data with uniform semantics; automatically constructing a cross-system call chain based on the request identifier and the like; secondly, extracting log features and comparing the log features with a normal behavior model to carry out anomaly detection, and generating an abnormal event; combining time, calling and dependency to construct an anomaly propagation path, and reversely backtracking and positioning an anomaly source; abnormal priority scores are calculated according to the propagation range, the influence degree and the like, and sorting and visual display are carried out; through unified semantic modeling and automatic association analysis of heterogeneous logs, accurate tracing and intelligent positioning of cross-service link anomalies are realized, the operation and maintenance efficiency is effectively improved, and the dependence on artificial experience is reduced.

The invention discloses a binary vulnerability data set expansion method and system based on cross-modal alignment, provides a hierarchical semantic fusion alignment framework, and remarkably improves the precision and expandability of binary vulnerability detection through a heterogeneous modal semantic bridging and transfer learning mechanism. A multi-modal semantic mapping channel is constructed by utilizing natural language interpretation and combining program structured analysis and constant anchor point coding, so that the semantic difference between a binary code and a source code is effectively bridged; a hierarchical attention mechanism enhances the fine-grained perception capability of cross-modal matching; furthermore, a migration framework driven by a vulnerability detection task is provided, binary vulnerability detection is mapped to a source code feature space through cross-modal alignment, a binary vulnerability data set is rapidly expanded by utilizing rich source code vulnerability data, and the data scale bottleneck is broken through.

The invention belongs to the field of natural language processing, particularly relates to an aviation safety accident report analysis method based on topic modeling and a word co-occurrence network, and aims to solve the problems that an existing topic modeling method is limited in understanding ability in aviation safety accident text analysis and cannot quantitatively reveal key causes. The method comprises the steps that a report text is acquired and preprocessed; based on a preset semantic fusion enhanced topic modeling engine keyword, obtaining a semantic fusion enhanced feature vector, and extracting a topic structure by adopting an improved deep embedding clustering model; and constructing an aviation safety accident word co-occurrence network based on a dynamic cosine similarity threshold value, performing network analysis, and determining key risk factors. According to the method, structured topic mining and deep learning semantic extraction are dynamically fused, and a network analysis method is combined, so that deep analysis and risk identification of the aviation safety accident report are realized, and the limitation of a traditional method on capturing text deep semantic information and subtle semantic difference is broken through.

The invention discloses a transnational trust modeling method fusing quantum state mapping and language phase difference, comprising the following steps: 1) estimating the site of each country based on the voting data of the united countries, and calculating the trust weight between countries through a weighting function in combination with the political ideal point distance and voting consistency of each country to obtain the initial quantized value of transnational trust; 2) mapping the initial trust value to a quantum state on a Bloch ball, applying phase perturbation according to semantic difference reported by a core country media, and simulating trust interaction between countries under discourse difference in combination with a quantum game mechanism so as to obtain a trust value mapped back to a classical space; and 3) carrying out iterative updating on the trust value by adopting a social influence model, dynamically describing an evolution process of the trust relationship between the countries, and outputting a time sequence change result of the trust network between the countries. According to the method, by fusing the international voting behavior and the transnational utterance difference, dynamic and quantifiable modeling of the inter-country trust relationship is realized, and an innovative methodological support is provided for international relationship research, transnational strategic game analysis and global risk early warning.

The invention relates to the technical field of data analysis, in particular to a method for automatically extracting and analyzing key information of massive credit short messages. The method comprises the following steps: replacing specific information in a short message text with a preset semantic tag, and performing word segmentation and vectorization processing on the short message text to obtain a structured text representation; according to each segmented word in the semantic unit, a corresponding preset semantic tag and a template semantic anchor point, determining a semantic anchor point feature index; screening out a target semantic anchor point; for each semantic analysis window, determining semantic description strength according to the number of target semantic anchor points in the window and semantic anchor point feature indexes; determining template similarity according to semantic difference characteristics of a target semantic anchor point and a template semantic anchor point in the window; according to the semantic description strength and the template similarity, determining the regional attention weight of the window; and inputting the word vector of each text slice and the attention weight of the corresponding region into a pre-training language model, and outputting text key information, thereby improving the accuracy of the key information.

The invention provides a resource recommendation method, a method and device for training a deep learning model and an intelligent agent, and relates to the technical field of artificial intelligence, in particular to the technical fields of resource recommendation, intelligent search, big data and the like. The resource recommendation method comprises the steps of obtaining a resource content feature sequence for a target object; fusing a plurality of resource content features in the resource content feature sequence based on a window attention mechanism to obtain a plurality of resource fusion features; and determining a target resource from the candidate resources based on the plurality of resource fusion features, and recommending the target resource to the target object, a candidate topic of the candidate resource and an initial topic for the resource content feature sequence satisfying a semantic difference condition.

The invention provides a natural image matting method and system based on text and boundary information aggregation, and relates to the technical field of image processing. Splicing the original color image and the corresponding ternary image, extracting initial features, and obtaining enhanced fusion features through multi-scale Laplacian high-frequency extraction and cosine similarity weighted fusion; based on the enhanced fusion feature and the ternary image, generating a gated ternary fusion feature fused with priori knowledge through a trans-attention mechanism; global coding modeling is carried out on the gated three-value fusion features to obtain deep features; text prompt and multi-scale boundary information are introduced based on deep features, adaptive up-sampling is guided through a cross-attention mechanism, semantic difference consistency constraint is adopted between decoding layers, consistency constraint is implemented from pixel appearance, high-level semantics and color component dimensions, and finally a transparency image is output through a prediction header to obtain an image matting result. And the fidelity and the boundary accuracy of high-frequency details in a matting result are effectively improved.

The invention relates to the technical field of natural language processing, in particular to a long text matching method and system based on key information and difference characteristics, and solves the problems that in the prior art, key information is dispersed due to noise interference in long text processing, and semantics of words or phrases in long text processing are more fuzzy and diversified. According to the method, a long text is preprocessed to obtain a test set sentence pair, a training set sentence pair and a verification set sentence pair, then a training model is obtained through sentence-level information entropy screening, word-level dynamic filtering, semantic difference enhancement and adaptive feature fusion, and finally after verification is conducted through the verification set sentence pair, the test set sentence pair predicts and outputs a result. The system comprises a long text matching system data preprocessing unit, a long text matching system function module training unit and a long text matching system function module result output unit. Key information is extracted for learning in long text matching, a large amount of computing power does not need to be consumed, and the matching effect is improved.