83 results about "Similarity analysis" patented technology

The invention provides a wrong question book intelligent learning system and method based on OCR and similarity analysis. The method comprises the following steps: S1, carrying out wrong question picture preprocessing and feature extraction; s2, multi-dimensional similarity calculation is associated with knowledge points; s3, knowledge point error rate statistics and error pattern analysis; and S4, generating and pushing personalized exercises. According to the method, intelligence and individuation of error question management are realized through multi-technology fusion, the bottleneck of current error question management is broken through, and the learning efficiency and the knowledge mastering depth are improved.

The invention belongs to the technical field of bid and bid behavior monitoring, and particularly discloses a bid and bid behavior monitoring management platform based on artificial intelligence, and the bid and bid behavior monitoring management platform adopts a chart normalization preprocessing technology, carries out the multi-dimensional similarity analysis of text content, chart features and image-text matching degree, and combines the order of images and texts with a recombination comparison mode. Omnibearing detection of the bidding document is realized, and the recognition rate of chart tampering and content recombination type bidding behaviors is effectively improved; based on a dynamic weight distribution mechanism of a project type, accurate risk assessment of different importance contents is realized through intelligent segmentation and differential weight setting of bidding document unit bodies; the multi-dimensional risk fusion analysis relation is constructed, the risk index is generated through comprehensive evaluation, and graded early warning is triggered, so that targeted monitoring on different structural weight markers is ensured, the detection breadth and depth are remarkably improved, and the problem of stiffness of a multi-modal analysis evaluation standard in the prior art is solved.

The invention relates to an electric power project demand similarity analysis method based on a rearrangement model and retrieval enhancement generation, and belongs to the technical field of computer science, artificial intelligence and natural language processing. The method comprises the following steps: constructing a multi-modal knowledge base; after user query is received, self-adaptive query expansion is carried out through a large language model, preliminary recall is carried out by adopting a mixed search strategy, secondary accurate scoring and reordering are carried out on candidate lists by utilizing a rearrangement model of a cross encoder framework, and a final result is generated in combination with an interpretability analysis technology. And finally, driving the large language model to generate a multi-dimensional intelligent analysis report through chained thinking prompt. According to the method, the problem of analysis deviation caused by low retrieval quality in a traditional RAG method is solved, the transparency of the system is enhanced through interpretability analysis, and the finally generated multi-dimensional analysis report can provide unprecedented decision support for project management.

The invention belongs to the technical field of data analysis, and provides an abnormal data analysis method and system in a financial fast report scene based on artificial intelligence, and the method comprises the steps: comparing obtained target financial data with a pre-configured threshold value through employing a three-dimensional threshold value model, judging whether there is abnormal data or not according to a comparison result, and carrying out the analysis of the abnormal data in a financial fast report scene. The threshold value of the three-dimensional threshold value model is configured by integrating an industry financial benchmark, historical financial data of the enterprise and financial data of the enterprise in the current year; if the abnormal data exists, retrieving data associated with the abnormal data by using the updated knowledge graph, and identifying a business event having correlation with the abnormal data according to the retrieved data; and performing similarity analysis, statistical significance analysis and anomaly classification analysis on the abnormal data and the business event, and performing weighted fusion to obtain a final analysis result. The accuracy of financial abnormal data analysis in the prior art is improved.

The invention discloses a medium surface microstructure detection device and method based on a spectrum technology, and relates to the technical field of structure detection, and the device comprises a partition division module which is used for dividing a target medium surface to obtain a medium partition sequence, and collecting a reflection spectrum of a surface microstructure to obtain a reflection spectrum sequence; the quality identification module is used for carrying out quality identification on the target medium to obtain a quality coefficient sequence and configuring abnormal residue identification precision; the exception identification module is used for calling exception residue identification resources and carrying out exception residue identification on the reflection spectrum sequence to obtain an exception residue parameter sequence; and the result processing module is used for performing processing to obtain abnormal residual distribution, performing similarity analysis on the abnormal residual distribution and historical abnormal residual distribution, obtaining identification confidence, combining the quality coefficient sequence and the abnormal residual parameter sequence, and performing calculation to obtain a medium quality parameter as a detection result. The technical problem that the medium surface detection effect is poor in the prior art is solved.

A device may process data associated with a first entity and a second entity, using one or more machine learning techniques, to identify a set of trends associated with a set of documents. The device may receive new documents associated with the first entity and the second entity. The device may generate, based on the set of trends, a first set of exceptions indicating that a new document is associated with a first type of error. The device may generate, using a similarity analysis technique, a second set of exceptions indicating that a new document is associated with a second type of error. The device may communicate with one or more systems to perform one or more actions associated with correction or prevention of processing errors relating to the new document based a claim relating to the first set of exceptions or the second set of exceptions.

The invention relates to a multi-dimensional feature-based case series-parallel analysis method and device, computer equipment and a storage medium. The method comprises the following steps: performing feature extraction processing on denoised sample data of each case sample to obtain a case feature set of the corresponding case sample; performing vector construction processing according to the case feature set of each case sample to obtain a multi-dimensional feature vector of the corresponding case sample, and performing calculation according to the multi-dimensional feature vector of each case sample to obtain a reference multi-dimensional feature vector; performing similarity analysis according to the multi-dimensional feature vector of each case sample and the reference multi-dimensional feature vector to obtain case similarity of each case sample; wherein the case similarity is the similarity between the multi-dimensional feature vector of the corresponding case sample and the reference multi-dimensional feature vector; in response to the existence of the target case similarity, recording a case identity identification number of a case sample corresponding to the target case similarity to a case string, and analyzing a statistical result; by adopting the method, the efficiency and the accuracy can be improved.

The invention relates to the technical field of ecological environment monitoring, in particular to an environment assessment method and device based on carbon footprint and regional ecological data. The method comprises the following steps: acquiring carbon footprint data and regional ecological data of a target region; performing dimension alignment on the carbon footprint data and the regional ecological data to obtain an associated data set associated with a preset grid of the target region; and based on the associated data set, through at least one of a time sequence similarity analysis strategy, a spatial distribution overlapping degree analysis strategy and a causal relationship influence analysis strategy, determining an environment assessment result of the target area. Compared with the prior art, a manual evaluation mode is replaced, and the analysis efficiency is improved. And on the other hand, the carbon footprint data and the regional ecological data are aligned according to the preset grids of the region, so that the problem of non-uniform comparison dimensions and indexes is solved. And meanwhile, three analysis strategies are provided for reflecting the internal relation of the two types of data, so that the data analysis precision is improved.

The invention relates to the technical field of network security, and discloses a program-level code similarity analysis method and device based on semantic slicing, and the method comprises the steps: constructing a program-level target function call graph based on a target source code; matching the target function call graph with a standard function call graph to obtain a plurality of isomorphic sub-graphs; performing semantic slicing on each function node in the first sub-graph and the second sub-graph to obtain a first slicing result of the first sub-graph and a second slicing result of the second sub-graph; and evaluating the slice similarity between each function in the first slice result and each function in the second slice result, and determining the similarity of the target source code according to the slice similarity. According to the method, the user-defined function call graph for eliminating system function interference is constructed, an isomorphic sub-graph matching algorithm is combined, multi-dimensional features are comprehensively considered, the accuracy of similarity calculation is improved, effective support is provided for practical application scenes such as code reconstruction and security audit, and the guarantee of code quality and security is improved.

The application provides a tunnel positioning method, device, equipment, medium and product, the method comprises the following steps: obtaining a plurality of network parameters reported by a user terminal when the user terminal is at a to-be-positioned point in a tunnel, predicted position data of the to-be-positioned point output by an inertial navigation system, and tunnel position data; based on the plurality of network parameters and the tunnel position data, a plurality of first MR fingerprint data are matched from an MR fingerprint library; similarity analysis is performed based on the plurality of first MR fingerprint data and the predicted position data, and MR fingerprint positioning data of the to-be-positioned point is obtained; and position data correction is performed based on the MR fingerprint positioning data and the predicted position data, and actual positioning data of the to-be-positioned point is obtained. The tunnel positioning method provided by the application can accurately calculate the positioning data of a user in a tunnel environment through an algorithm without the need to separately deploy a positioning device in the tunnel environment and without the need to be limited by hardware devices, so that the flexibility is improved while the accuracy is ensured.

The invention discloses a science and technology project document duplicate checking method and device based on multi-level abstract generation and a medium, belongs to the technical field of document duplicate checking, and solves the problem of how to improve the reliability and accuracy of science and technology project document duplicate checking. The method comprises the following steps: firstly, carrying out structured analysis on a project document to be subjected to duplicate checking, introducing a fine-tuned abstract generation model to obtain a multi-level abstract of a project to be subjected to duplicate checking, and extracting a feature vector of the multi-level abstract through a fine-tuned and trained text feature extraction model; secondly, performing similarity search in a vector database based on feature vectors corresponding to the dimensions of the main abstract, further calculating cosine similarities based on the dimensions of the subordinate abstract, and sequencing the weighted similarities to obtain a similarity rank; and finally, according to the structured analysis information and the similarity ranking result, LLM similarity analysis is performed by using a large language model, a document duplicate checking report is formed, the originality of the science and technology project is ensured, repeated subsidization is avoided, and the method has important practical application value for science and technology project management.

The invention discloses an object recognition method and device, equipment and a storage medium, and relates to the technical field of machine learning, and the method comprises the steps: carrying out the feature similarity analysis of a target object feature of a to-be-recognized object and a class index feature group corresponding to a target object class, and obtaining target feature similarity data; the category index feature group is a clustering center feature corresponding to the positive sample object feature; the positive sample object is a sample object which is screened based on a feature density clustering result of a plurality of original sample objects and belongs to a target object category; based on the target feature similarity data and a preset feature similarity threshold, determining object category indication information of the to-be-recognized object; the preset feature similarity threshold value is obtained by performing threshold value adaptive adjustment based on a preset identification index threshold value and the sample feature similarity data; the sample feature similar data represents the feature similarity condition between the positive sample object feature of the positive sample object and the category index feature group. By using the scheme of the invention, accurate identification of a new category can be rapidly realized.

The present invention belongs to the field of Internet of Things security technology, and discloses a firmware password library detection method and device based on similarity analysis. The method includes: constructing a public password library data set, obtaining the binary file of the public password library data set as the data set binary file, and unpacking the firmware to be detected to obtain the binary file of the firmware as the binary file to be analyzed; performing file similarity analysis on the binary file to be analyzed and the data set binary file to identify the public password library; in the files where the public password library is not identified, clustering according to the file similarity to identify the private password library. In the process of performing similarity analysis, the method of the present invention not only utilizes the features inside the function, but also uses the sequence order between functions as the basis for similarity recognition, so as to be able to accurately identify the password library in the Internet of Things firmware and restore the APIs of all functions in the library, which greatly facilitates the subsequent security analysis work.

The invention relates to a historical case similarity-based quality evaluation method, which belongs to the field of software development, and comprises the following steps of: screening out a prior index and a posterior index; collecting data of each stage in the research and development process of each software historical version as historical cases; the specific numerical value of each index is counted; obtaining quality problem source data of the historical cases; taking the historical cases with the labels as input samples, performing model training by adopting an SVM classification algorithm, and constructing a quality problem classification model; constructing a prior index vector matrix for each historical case and the newly-added case, and calculating the similarity between the newly-added case and each historical case; automatically classifying the newly added cases and setting labels; calculating a risk value of each similar case; and averaging the risk values of all the similar cases to obtain a risk coefficient value of the newly added case. According to the method, the influence of manual subjective deviation on index interpretation can be avoided, and an objective quality evaluation standard is established through quantitative similarity analysis.

The application discloses a role copyright infringement detection method and system based on visual and semantic features, the method comprises the following steps: S1, role target separation and feature extraction: input the content to be detected, separate the role target therein, and extract the role features, the role features comprising at least one of visual features, semantic features and time sequence features; S2, role feature correspondence and matching: the extracted role features are matched with the existing role features in the copyright content library; S3, role feature similarity analysis: when the similarity between the extracted role and the matched role exceeds a preset threshold, the role is determined as a potential infringing role. The application takes the visual and semantic features of the role as the core, constructs a whole-process detection framework from target separation to feature matching and then to similarity analysis, can effectively solve the low-efficiency problem of the prior art which depends on the manual comparison method, provides an automatic and intelligent solution for the role copyright infringement identification of large-scale digital content, and significantly improves the detection efficiency and the scientificity of the identification.

The present invention discloses a device and method for detecting the microstructure of a medium surface based on spectral technology, which relates to the field of structural detection technology. The device includes: a partitioning module for partitioning the surface of a target medium to obtain a medium partition sequence, collecting the reflectance spectrum of the surface microstructure to obtain a reflectance spectrum sequence; a quality identification module for performing quality identification of the target medium to obtain a quality coefficient sequence, and configuring the abnormal residue identification accuracy; an abnormal identification module for calling abnormal residue identification resources, performing abnormal residue identification on the reflectance spectrum sequence to obtain an abnormal residue parameter sequence; and a result processing module for processing the obtained abnormal residue distribution, performing similarity analysis with historical abnormal residue distributions, obtaining identification confidence, and combining the quality coefficient sequence and the abnormal residue parameter sequence to calculate and obtain the medium quality parameter as the detection result. The present invention solves the technical problem of poor medium surface monitoring and detection effect in the prior art.

The invention relates to the technical field of data management, in particular to an e-commerce business information data management platform based on big data, which comprises an e-commerce business information classification module for accurately classifying collected e-commerce business information data; and the e-commerce business information analysis management module is used for carrying out integrity verification on input data in the basic analysis layer in the data analysis model based on the constructed integrity verification control group of the basic analysis layer, and carrying out deep analysis management by utilizing a graph structure analysis model. According to the method, the graph structure model with the feature vectors as the nodes and the behavior chains as the edges is constructed, the chain type conversion logic of the e-commerce users from browsing, behaviors to purchasing can be completely captured, the user behavior mode and the commodity relation are effectively recognized, and the limitation of a traditional flat data analysis mode is broken through; and multiple rounds of similarity analysis are performed on the residual data in the classification process, and unmatched data are regarded as abnormal data and are eliminated, so that the overall data quality is improved from the source.

A large language model-oriented multi-sample attack detection and defense method relates to the field of artificial intelligence security. A multi-level text screening and intervention mechanism is constructed to reduce the impact of multi-sample attacks through input content deduplication, format adjustment and context disturbance. Similarity analysis is performed on the text to identify highly repetitive examples, which are screened using an efficient matching strategy. The semantic features of the text are mapped using a pre-trained model, and historical data are compared to determine whether there is a potential inducement risk. The multi-round dialogue structure is analyzed, and the user and system interaction mode is extracted and analyzed to dynamically adjust the detection rules and improve the adaptability to different format inducement attacks. For high-risk text, deletion and truncation strategies are adopted, and for some medium-risk text, the order is adjusted or interference information is inserted to reduce its impact. The safety of LLM in complex interactive environments is improved, and the generation of non-compliant content is reduced, which is efficient, low in false positives and scalable.

The present invention belongs to the technical field of data processing of underlying sensors in energy consumption monitoring systems, and provides a method for real-time reconstruction and online repair of missing data of sensors in an energy monitoring system, comprising the following steps: S1, establishing a system model and a digital model of the energy monitoring system based on its design principle and operation mechanism; S2, performing similarity analysis on the data; S3, constructing a filled data set based on the maximum expectation method and the maximum likelihood estimation algorithm; S4, Bayesian data reconstruction based on the MCMC algorithm. Beneficial effects of the present invention: Based on data and models, the present invention can reconstruct and online repair missing data of one or multiple types of sensors in the energy monitoring system in real time or simultaneously, which can ensure the accuracy of the estimated value, and has high filling efficiency and large filling dimension, while achieving the accuracy of historical data and the integrity of the database, which is of great significance to the improvement of the intelligent energy monitoring system.

The application discloses a sparse feature selection and index dynamic configuration method for water ecological environment monitoring, and relates to the technical field of water ecological environment monitoring. The method comprises the following steps: collecting water ecological environment data, projecting to a feature domain and a response domain through a feature mapping algorithm, and establishing a double-domain correlation; calculating a sensitive coefficient to screen key features, eliminating redundancy through similarity analysis to form a core feature set; dividing the indexes into benchmark and floating indexes, dynamically adjusting the weight and frequency of the floating indexes according to a balance threshold; constructing a dynamic iteration model to realize dynamic matching of features and indexes; generating a sparse feature subset suitable for different spatial scales through multi-resolution fusion; and verifying stability and triggering linkage optimization through a disturbance scenario simulation. The application solves the problems of feature redundancy, index rigidity and poor scale adaptation in traditional monitoring, improves the monitoring accuracy, efficiency and stability, and is suitable for complex basin multi-scenario monitoring requirements.

The invention discloses a multi-scale multi-view target similarity analysis method based on deep learning, and relates to the technical field of information perception and recognition. The method comprises the following steps: preprocessing existing paired images to obtain model input data; the processed data are respectively input into a ResNet50-based feature extraction channel and a pre-trained visual large model-based feature coding channel, while feature information of an image pair is fully learned, robustness features obtained by direct reasoning of the large model are effectively utilized, and feature fusion is performed, so that a dual-channel feature extraction structure is constructed; a group of optimal model parameters are obtained through optimization of a certain number of iterations; and reasoning the trained model to obtain a similarity result of the input paired images. According to the method, the model recognition precision can be effectively improved under the condition that computing resources are not excessively consumed by utilizing the robustness features obtained by reasoning the pre-trained large model while the feature information of the image pair is fully learned.

The invention relates to a computer-implemented method for inspecting or operating a first product (TD1), the following steps being carried out: a) transmitting a provided global model (GM) of the first product (TD1) or the operation thereof from the server (S) to the first client (C1) and second client (C2), b) capturing a first data set (DS1) of the first product (TD1) or the operation thereof with first product parameters of the first product (TD1) as respective first subsets (C1.1-C1.3), and generating and training respective first local models (LM1.1-LM1.3) with the respective first subsets (C1.1-C1.3), c) capturing a respective second data set (DS2) of the respective second product (TD2) or the operation thereof with respective second product parameters of the respective second product (TD2) as respective second subsets (C2.1-C2.3), and generating and training respective second local models (LM2.1-LM2.3) with the respective second subsets (C2.1-C2.3), d) performing a similarity analysis between models of the first and the second local models (LM1.1-LM1.3, LM2.1-LM2.3), and assigning these similar models to a common model group (G1-G3), e) training a global detail model (M1) on the basis of the model of the common model group (G1-G3) and transmitting same to the first client (C1), f) applying the detail model (M1) to the first product (D1) for inspection or operation thereof, by the first client (C1).

The present application relates to the system failure analysis in the security field, provide a kind of event similarity analysis method based on factor, for the numerous events involved in the system failure evolution process, a kind of event similarity analysis method based on factor is presented;Method is based on influencing factor and change, event and event failure probability, involves improving pearson similarity, information entropy and system failure evolution process theory;Firstly, the feasibility of event similarity based on factor analysis is studied;Secondly, the event similarity analysis method is established, including event failure probability similarity, influencing factor similarity, probability factor similarity, event failure probability entropy similarity and event similarity;Thirdly, the event classification method is proposed for the different event similarity;For the event classification based on event similarity.The original event set is simplified by the above method, so that the system failure evolution process reduces the complexity of construction.

A method may include obtaining well data regarding a geological region of interest. The method may further include obtaining a geological process model for the geological region of interest. The method may further include determining a simulation region of the geological process model based on a template size and a simulation grid. The method may further include determining a simulation path through the simulation region. The method may further include determining a geological event associated with a predetermined geological facies based on the well data. The method may further include determining a replicate event based on the geological event, a similarity analysis, and using the simulation path. The method may further include assigning the predetermined geological facies to a cell location of the replicate event in the simulation region. The method may further include updating the geological process model to produce an updated geological process model.

The present application relates to the technical field of livestock data processing, in particular to a livestock and veterinary breeding data analysis system, which comprises: a state transition analysis module collects data of the breeding cycle of female livestock, integrates the data of all female livestock to construct a transition probability matrix, predicts the future state transition path of female livestock according to the transition probability matrix, and generates a female livestock state transition evaluation result. In the present application, by analyzing the data of each state of the breeding cycle of female livestock, the state change path of female livestock at different stages is determined, accurate prediction and quantitative evaluation of the state in the breeding cycle are realized, and the formulation of the breeding plan is more accurate. By evaluating the residence time and transition trend of each stage, the best breeding time is predicted, and the overall breeding efficiency is improved. Based on the similarity analysis of behavior data such as activity frequency, body temperature and number of reproductions, individual female livestock is divided into different behavior groups, and more targeted reproduction management strategies are formulated for different groups.

The invention provides a time sequence similarity analysis method and device for sensor data in a wafer processing process, and the method comprises the steps: obtaining time sequence data collected by a sensor corresponding to a target parameter in the wafer processing process; based on the time sequence data, respectively determining a reference sequence and at least one comparison sequence; respectively carrying out dimensionless and normalized data preprocessing on the reference sequence and the at least one comparison sequence; for each comparison sequence, calculating a correlation coefficient of each data sampling point at the same position between the reference sequence and the comparison sequence based on a grey correlation analysis algorithm; and based on the correlation coefficient of each data sampling point, obtaining a correlation degree for quantitatively representing the correlation between the reference sequence and each comparison sequence. According to the method, the FDC system and the grey correlation analysis algorithm are combined, time sequence analysis can be automatically carried out on the time sequence data collected by the sensor, and the analysis efficiency and quality are improved.

The application discloses an evaluation method and an evaluation device for the stealth property of a columnar grating. The evaluation method comprises the following steps: firstly, calculating the theoretical critical stealth object distance of an object to be observed through a columnar grating with preset grating attribute parameters according to a theoretical critical stealth object distance formula of the columnar grating; secondly, changing the position of the object to be observed within a preset object distance range and acquiring images formed after the object to be observed passes through the columnar grating with the preset grating attribute parameters in real time; and finally, performing similarity analysis on the images formed after the object to be observed passes through the columnar grating with the preset grating attribute parameters and standard images to evaluate the stealth property of the columnar grating with the preset grating attribute parameters. The stealth property of the columnar grating with the preset grating attribute parameters can be evaluated clearly, accurately, quickly and intuitively by using the method, and the method has good experimental demonstration effect and measurement function, and has important guiding significance for designing and manufacturing the columnar grating with good optical stealth effect.

The invention discloses a hydropower station structural plane analysis method and system based on prompt learning and a medium. Relates to the technical field of engineering geology. The method comprises the following steps: acquiring multi-modal basic data of a structural surface of a hydropower station, and preprocessing the multi-modal basic data; inputting the preprocessed multi-modal basic data into the constructed structural surface analysis model for classification and similarity analysis; the structural surface analysis model performs cross-modal mapping of multi-modal basic data by sharing an embedding space, and introduces a prompt learning mechanism to perform similarity analysis to obtain a hydropower station structural surface classification result and a similarity result; according to the scheme, a multi-modal AI framework and domain knowledge driving method is fused, and cross-modal feature fusion of multi-modal data such as texts, images and spaces is realized through a multi-modal large model and a prompt learning technology; and meanwhile, engineering specifications are coded into priori knowledge of the model by utilizing a prompt learning template, so that the integrating degree of an output result and an industrial standard is enhanced.

The present application discloses a method, device, equipment and storage medium for assigning values to cross-system data objects, relating to the technical field of data processing. The method includes: when an object assignment instruction is recognized, obtaining a source object from a source business system and an object to be assigned from a target business system; the source business system and the target business system are different systems for processing the same target business; performing a field name consistency check and a data type consistency check on the source object and the object to be assigned; when it is recognized that both the field name consistency check and the data type consistency check fail, performing a semantic similarity analysis on the metadata of the source object and the object to be assigned to obtain a semantic similarity analysis result; and assigning a value to the object to be assigned according to the semantic similarity analysis result and the source object. The above technical solution helps to ensure the efficiency and accuracy of value assignment on the basis of realizing automatic value assignment.

This invention belongs to the field of garment production traceability technology. It provides a garment production quality traceability system and method, comprising: first, performing similarity analysis on traceability data of small batches affiliated with large-volume orders within a factory to identify highly similar affiliated groups; then, integrating similar non-traceability data of small batches and large-volume orders within each group into data pairs, calculating Euclidean distance, and ranking them from low to high based on distance; next, combining historical traceability records of similar small batch orders, counting the number of anomalies in various types of non-traceability data, and ranking traceability anomalies from high to low based on the number of anomalies; finally, determining the traceability data to be added based on the two ranking results. If the top-ranked data is the same, it is directly added; otherwise, the minimum value is selected from the sum of the rankings and added to the small batch order traceability data. This effectively avoids confusion between small batch and large-volume order traceability data, reduces the traceability error rate, and improves the efficiency of tracing quality issues in small batch orders.