50 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 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 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 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 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 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 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 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 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.

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.

The invention discloses an unsteady flow field characteristic and disturbance source matching identification method under multi-factor influence, and belongs to the technical field of unsteady flow field processing, and the method comprises the following steps: S1, carrying out simulation calculation according to the working environment of a turbine, and obtaining unsteady flow field data; s2, determining a mode corresponding to a disturbance source according to the unsteady flow field data; and S3, according to the mode corresponding to the disturbance source, determining the influence degree of the disturbance source on the unsteady flow field characteristics. According to the method, the unsteady flow field features and the disturbance sources can be matched and recognized, mutually orthogonal modes influenced by the disturbance sources are obtained through mode decomposition, and the influence effect of each disturbance source on the unsteady flow field features is accurately evaluated in combination with similarity analysis; and the application scene is not limited to an unsteady flow field of the turbine.

The application discloses a target and background spatio-temporal spectrum similarity analysis system and method, which comprises a spectrum pretreatment module, a spectrum feature extraction module, a spatio-temporal variation rule analysis module and a similarity analysis module; target and background spectra are transmitted to the spectrum feature extraction module after being processed by the spectrum pretreatment module; the spectrum feature extraction module extracts features of the spectra and quantifies differences, and then transmits the features to the spatio-temporal variation rule analysis module; the spatio-temporal variation rule analysis module analyzes the spectrum feature differences with the spatio-temporal variation rule, and transmits the variation rule to the similarity analysis module; the similarity analysis module combines the spectrum feature differences with the spatio-temporal variation rule, assigns weights to each feature, and comprehensively obtains a similarity value to grade the spectrum similarity. The application considers the influence of the time dimension and the space dimension on the spectrum features, combines the spectrum features with the variation rule to assign weights, and realizes the expansion of traditional static spectrum features to the analysis system and method of a dynamic scene.

A query is received which is to be input into a machine learning model (or other artificial intelligence model). Thereafter, a plurality of historical queries of the machine learning model meeting first criteria relative to the query is determined using a first distance-based similarity analysis technique. Each of the historical queries have a known output by the machine learning model. An output of the machine learning model responsive to query is received. Next, it is determined, using a second distance-based similarity analysis technique, whether the output meets second criteria relative to each of the known outputs corresponding to the historical queries. This determination characterizes whether the query is likely to cause the machine learning model to behave in an undesired manner and can be provided to a consuming application or process. Related apparatus, systems, and techniques are also described.

The invention relates to the technical field of judicial information matching, and discloses a judicial case intelligent matching system based on multi-modal data processing, and the system comprises a judicial case extraction module which is used for extracting to-be-matched judicial case data; the text acquisition module is used for inputting the extracted judicial case data into a multi-modal data model for feature tag extraction and similarity analysis to obtain similar case texts; the correlation degree matching module is used for identifying the correlation degree between the structured data in the similar case texts, judging the result of the correlation degree based on a convergence criterion, and matching the case texts according to the judgment result; and the visual display module is used for building a visual billboard. According to the method, high-precision similar case retrieval and matching on core information such as case facts, criminal names and judgment results and elements such as regions, time and parties are realized, and the accuracy and the intelligent degree of case comparison are improved.

The invention discloses a large model context error management method and system based on regression testing, belongs to the technical field of artificial intelligence, and aims to solve the technical problems of repeated errors, lack of error learning ability and imperfect error prevention mechanism existing in a large model at present. Comprising the following steps: constructing an error history library based on a triple data structure corresponding to historical error data; performing similarity analysis on the current user input and historical user input in an error historical library; generating a test case based on a historical user input similar to the current user input; optimizing the current user input to obtain optimized user input; and establishing an error reduction rate evaluation index, comparing error occurrence rates before and after user input optimization through an A / B test framework, and adaptively optimizing a similarity judgment standard, a weight parameter and an error avoidance instruction template based on a feedback result.

The present application belongs to the technical field of education data mining, and discloses a test question implicit knowledge attribute association mining method and system based on attribute exploration and a related test question pushing method and system, which comprises the following steps: constructing a formal background of a student's error question set; preprocessing a student's question answer record source data, and then filtering out the student's error answer information in combination with a knowledge attribute set contained in the labeled question; exploring the current formal background by using an attribute exploration knowledge attribute mining algorithm to obtain a knowledge attribute implication association set and a non-redundant test question set through any student error answer information; calculating the similarity between test questions by using concept lattice similarity analysis, finding questions containing similar knowledge points to the practice wrong questions, and selecting a number of related test questions meeting a threshold to push. The present application solves the drawbacks caused by the missing attributes in the students' fuzzy judgment in the traditional analysis mode, and can provide targeted test question recommendations for students.

The invention discloses a field similarity analysis method, a table similarity analysis method and a storage medium, and relates to the technical field of data processing.The method comprises the steps that a first field and a second field with the similarity to be analyzed are obtained, a pre-constructed processing link weight graph is obtained, the processing link weight graph comprises field nodes and weight connection edges, and the field nodes are connected with the weight connection edges; field nodes connected with the weight connection edges have an operation processing relation, and the edge weights of the weight connection edges are set based on the operation processing relation; identifying a processing path from the first field to the second field in the processing link weight map; the field similarity between the first field and the second field is determined according to the processing path, and the field similarity is in positive correlation with the edge weight of the weight connection edge in the processing path. According to the method and the device, efficient and high-accuracy field similarity analysis is realized.

This invention discloses a ResNet-based lightning-three-dimensional surrogate reflectance inversion method, belonging to applied meteorological technology. The method includes acquiring lightning location observation data, screening, quality control, and classification of lightning location observation records, determining a grid, statistically analyzing data at grid points, processing the statistical data, calculating spatial, temporal, and spatiotemporal coefficients, sorting the lightning location observation records for each grid point, processing the sorted lightning location observation records and other data to obtain the maximum surrogate reflectance, and obtaining the three-dimensional surrogate reflectance by similarity analysis of the maximum surrogate reflectance and its surrounding atmospheric environmental characteristics, thus achieving the inversion of the three-dimensional surrogate reflectance. This invention provides a more accurate inversion of the three-dimensional surrogate reflectance and effectively solves the problem of insufficient monitoring and early warning capabilities for severe convective weather in radar detection blind zones.

The application relates to a product recommendation method, a product recommendation device and a storage medium. The product recommendation method comprises the following steps: obtaining historical data of a product, and constructing a first data matrix according to the historical data; the first data matrix comprises function points supported by the product; performing similarity analysis on the first data matrix, and merging function points with a similarity meeting a set threshold into a first product function group according to a similarity analysis result; calculating information gain of the first product function group; selecting the first product function group according to the information gain, obtaining a second product function group; calculating a product recommendation probability of the second product function group, and recommending the product according to the product recommendation probability. Through the application, the calculation amount of product data analysis in a post-processing step is greatly reduced, and the calculation efficiency is improved.

The application discloses a trajectory similarity analysis method based on trajectory compression and clustering. Firstly, N reference trajectories are compressed, and some trajectory data is removed within a certain error range, so as to facilitate subsequent processing and analysis. Secondly, the range of trajectory segments participating in calculation is reduced by intercepting common trajectories and angle pre-judgment, and the reference trajectory data is clustered offline. When calculating the similarity matrix, the similarity of each two trajectories only needs to be calculated once within the error range to reduce the complexity. Finally, the representative trajectory of each cluster is found, the similarity of the to-be-matched trajectory and the representative trajectory of each cluster is calculated, and whether the to-be-matched trajectory exists in the reference trajectory database is judged. If the to-be-matched trajectory exists in the reference trajectory database, the trajectory similarity of the to-be-matched trajectory and each reference trajectory of the corresponding cluster is calculated, so that it is determined that the actual trajectory is similar to which reference trajectory, and a similarity result is obtained.

A device and method for reviewing literature by using Latent Dirichlet Allocation (LDA) is proposed. The device may include a pre-processing unit extracting text data for modeling, and a modeling unit automatically classifying topics as many as a set number (K) and generating a probability distribution of the topics by literature and a probability distribution for words by topic. The device may also include a clustering unit updating the number (K) of the topics, an interest analysis unit confirming trends by topic over time, and a generality analysis unit quantitatively confirming a research scope of each specific topic. The device may further include a similarity analysis unit quantitatively confirming a similarity between the topics, a network analysis unit quantitatively confirming a correlation between the topics, and a display displaying the trends by topic over time, research scope of each specific topic, similarity between the topics, and correlation between the topics.

The invention belongs to the technical field of artificial intelligence, and relates to an artificial intelligence-based data analysis method, which comprises the following steps of: respectively carrying out optical flow field calculation on an input query frame and a reference frame to obtain a first optical flow field and a second optical flow field; respectively performing feature extraction on the query frame and the reference frame to obtain a query frame feature and a reference frame feature; generating an adaptive threshold based on the successive frames; processing the first optical flow field, the query frame feature and the adaptive threshold to obtain a first attention weight map and a first weighted map feature, and processing the second optical flow field, the query frame feature and the adaptive threshold to obtain a second attention weight map and a second weighted map feature; and performing similarity calculation on the first attention weight map, the first weighted map feature, the second attention weight map and the second weighted map feature to obtain background similarity score data. The method can be applied to image analysis scenes in the financial science and technology field and the digital medical field, and the processing accuracy of image background similarity analysis can be improved.

Embodiments are directed to a computer-implemented method of analyzing a land region that has been decomposed into a plurality of regular or irregular sub-regions. The computer-implemented method includes applying, using a processor system, a feature extraction process that extracts a set of sub-region environmental descriptors for each of the plurality of sub-regions. The processor system applies a similarity analysis to the set of sub-region environmental descriptors to generate groups of the plurality of sub-regions. The processor system creates a plurality of group-based graphs by encoding each of the groups into a corresponding group-based graph. A spatio-temporal neural network is used to train a model based at least in part on the plurality of group-based graphs.