19 results about "Text filtering" patented technology

The invention provides a Web-based intelligent image creation system and an implementation method thereof. The system comprises a Web application service module, an image creation module and an image creation module, the model service cluster comprises a Docker container deployment text-to-image generation model based on Kubernetes management, an image-to-image conversion model and a low-rank adaptive model LoRA; the distributed file storage system is used for storing a user-defined model, a training data set and a generated work and supporting multi-copy redundant backup; the integrated LoRA training module provides data set labeling, trigger word binding and training parameter optimization functions; the multi-modal image processing module has the functions of local redrawing, intelligent image expansion, super-resolution repair and style migration; and the dual-channel content security auditing module adopts BERT text filtering and YOLOv5 image recognition technologies. The technical defect that an existing AI generation tool cannot support a user-defined model and a dynamic extension service is overcome, and the whole-process closed-loop operation of a user from data labeling to model deployment is supported.

The invention discloses an intelligent customer opinion processing system based on multiple agents. A receiving module receives complaint information of a user and standardizes the complaint information. And the task distribution module realizes splitting of customer complaint tasks and intelligent agent adaptation for standardized customer complaint data through a multi-intelligent agent iterative learning framework. And the subtask execution module executes the received customer complaint task. And the feedback optimization module collects full-link indexes in real time, detects abnormal events by adopting a sliding time window, and optimizes the construction of a decision tree in the sub-task execution module. And the content filtering module is used for carrying out post-processing on output texts of the consultation agent and the risk identification agent by using a text filtering and optimizing framework based on an off-line large model. And the result storage module adopts a multi-level storage strategy to record a full-link operation log. According to the method, the problems of low efficiency, insufficient accuracy, weak dynamic adaptive capacity and the like of a traditional customer complaint processing method can be effectively solved.

The invention discloses an art image style migration method and device, equipment, a medium and a product, and relates to the field of image processing, and the method comprises the steps: obtaining a to-be-processed style picture and a to-be-processed content picture; performing style text filtering on the style picture to be processed to obtain a pure content description text; the style text filtering comprises style shift-out and picture-text distance comparison; creating a single pseudo word and expanding the single pseudo word based on a time step stage and a sub-attention layer to obtain a style pseudo word set; optimizing the style pseudo-word set by utilizing inversion of a diffusion model according to the pure content description text to obtain style representation of the style picture to be processed; and taking the style representation of the style picture to be processed as style guidance, determining content guidance according to the content picture, and determining a stylization result by utilizing denoising of the diffusion model. The stylization effect can be improved.

The invention relates to the technical field of data quality evaluation, in particular to an automatic quality evaluation method for a high-quality data set in the cultural field, which comprises the following steps: extracting semantic fragments and vector information, performing clustering analysis on a coverage structure, screening redundant samples, labeling image-text differences, and outputting a consistency evaluation result. According to the method, high-precision recognition of semantic coverage and structure density can be realized through structured extraction of semantic information and vector distribution measurement, and the discriminability of redundancy detection is enhanced through linkage screening of vector coincidence frequency and similarity threshold. The systematicness of image-text consistency marking is improved through an image-text position difference and semantic span combined marking mechanism, and it is ensured that redundant and deviated intersection samples can be accurately recognized through a synchronous contrast mechanism between text screening content and image-text difference marking results. The intensive clustering of the cultural semantic content in the vector space and the quantitative analysis capability of the structure deviation point are synchronously improved, and the comprehensive dimension and logic closure of the evaluation result are optimized.

This invention relates to the field of intelligent question answering technology, specifically to a RAG question answering system combining knowledge tracking and evidence annotation. The system includes a semantic parsing module, a sentence localization module, a knowledge path annotation module, an evidence indexing module, and an enhanced question answering module. This invention extracts explicit verb semantics, indicator entities, and time-normalized expressions through structured parsing of natural language input, ensuring structural uniformity and processing consistency of the input sentences. A tag-based text filtering strategy effectively limits the text scanning range, improving the accuracy and efficiency of indicator localization. It clearly extracts statistically significant causal links from the text, and enhances the cross-coverage of indicator fragments by imposing indicator combination constraints and labeling key sentences in path nodes, thereby improving the density and logical order consistency of indicator integration in the answer.

The invention discloses an RAG multi-hop question and answer method, device and system and a storage medium. The method comprises the steps of training a planning model and an information extraction model; analyzing the complex problem into a retrieval sub-task and a reasoning sub-task through the trained planning model; performing dual-path document retrieval by combining a cosine similarity algorithm and a relative boundary distance algorithm to obtain a main index result and an auxiliary retrieval result; performing parallel processing on the main index result and the auxiliary index result, and sequentially connecting the processing results to obtain a final retrieval result; wherein the parallel processing is that corresponding text segments are integrated based on themes and sentence-level text filtering is carried out; and processing the retrieval sub-task and the reasoning sub-task by using the trained information extraction model, and integrating all intermediate results to generate a final answer. Through the task decomposition planning and double-path index optimization technology, the interpretability of the reasoning process is improved, the question and answer accuracy and reliability are improved, and the method can be widely applied to complex multi-hop question and answer scenes.

A computer-implemented method for updating one or more inversed lists of a read / write graph database, each inversed list being associated with a trigram comprised in a literal value. The method including obtaining the dictionary encoder including a set of literal values and a set of IDs, each literal value being associated with an ID. The method further includes, for each literal value, identifying trigrams composing the literal value. The method further includes, for each identified trigram, and in a first data structure, updating a header of the inversed list associated with the identified trigram, and updating a designated block which includes a representation of the ID of the literal value including the trigram, thereby updating the inversed list associated to the identified trigram.

The present invention discloses a method for intelligent traffic event recognition based on a large traffic model using cross-modal retrieval. The method comprises the following steps: inputting an image of a traffic event to be recognized and its text description into a traffic event recognition network model; the traffic event recognition network model comprises: a visual feature extraction module for extracting visual features from the image; a text feature extraction module for extracting text features from the text description; a visual selective filtering module for selectively filtering irrelevant visual extraction features to obtain visual filtering features; a text selective filtering module for selectively filtering unnecessary text extraction features to obtain text filtering features; a cross-modal selective alignment module for cross-modally aligning visual and text extraction features to obtain visual alignment features and text alignment features; and a calculation module for fusing the visual filtering features and visual alignment features, as well as the text filtering features and text alignment features, to determine a recognition result. The present invention improves the accuracy and efficiency of retrieval and recognition.

An embodiment of the present invention provides a method and apparatus for identifying the position of a text area. The method includes: obtaining a frame image to be identified, where the frame image includes a text area; extracting the feature values of each pixel point in the frame image; calculating the gradient of each pixel point according to the feature values, and identifying the position of the frame image corresponding to the gradient peak among each gradient from the frame image; counting the number of pixel points corresponding to the gradient peak in the frame image position; and based on the comparison result between the number of pixel points and a preset threshold, confirming that the frame image position is the position of the text area. The embodiment of the present invention preliminarily determines the row position or column position of the text area, and then based on the comparison result between the number of pixel points corresponding to the row position or column position and the preset threshold, finally confirms that the frame image position is the position of the text area, which can provide a filtering basis for filtering the text in the subsequent text area, achieve the effect of narrowing the scope of text filtering, reducing the calculation amount of text filtering, and optimizing the text filtering effect.

This disclosure belongs to the field of nuclear power technology, specifically relating to a personalized question-and-answer method and apparatus for nuclear power management regulations. This disclosure aims to address the impact of user personalization characteristics on question-and-answer results, proposing a personalized question-and-answer method for nuclear power management regulations based on the collaboration of user characteristics and a large language model. The method of this disclosure sorts the retrieved text information based on a feature association module of user characteristics and semantic units. Then, the relevant text and questions are input into a large language model, which generates personalized answers based on the user's personalized characteristics. This allows for optimized text filtering by combining user characteristics and integrating retrieval enhancement generation with a large language model to generate personalized answers, more accurately meeting the needs of different users.

The present disclosure provides a multimodal feature fusion method and system for sentiment analysis, which includes: obtaining text data and its corresponding image data in social data; performing feature extraction on the text data and image data respectively to obtain text features and image features; based on the text features and image features, using a cross-modal attention mechanism to obtain interaction information between the two, and performing noise filtering on the interaction information through a gating mechanism to obtain text features after image filtering and image features after text filtering; splicing the obtained text features, image features, text features after image filtering, and image features after text filtering to obtain fused features; and based on the fused features, obtaining sentiment analysis results through a pre-built sentiment analysis model.

This invention relates to the field of text extraction, specifically to a deep learning-based system and method for extracting key text features. The system includes a text filtering module, a graph construction module, a deep learning module, a cluster feature module, and an input fusion module. The text filtering module merges text clusters, the graph construction module generates a stereo model, the deep learning module expands the model using clustered text, the cluster feature module prunes the stereo model, and the input fusion module captures new text features to generate fused text. This invention provides a visual interface for training complex text, enhances the machine's ability to analyze the coherence of long texts, improves the generalization ability of machine learning, is suitable for processing large amounts of complex, high-dimensional text data, reduces external data requirements, adapts to changes in data distribution, reduces the computational burden on the model, and achieves stable language frameworks and precise fusion of text features.

The invention discloses an LBS parallel judgment method, system and device and a storage medium, and the method comprises the steps: obtaining and carrying out the parallel matching of user position information and a business text, and carrying out the intersection processing of a first object list and a second object list obtained through matching, and obtaining a target object list; the matching of the user position information comprises the following steps: acquiring a first data set comprising an object identifier and a position identifier from a remote cache; starting a plurality of processing threads, fragmenting the first data set, and distributing the fragmented first data set to the plurality of processing threads; each processing thread reads geographical range information from a near cache composed of an out-of-heap memory and a shared disk according to the object identifier and the position identifier, and matches the user position information with the geographical range information to determine a matched object; and summarizing the matching objects of the plurality of processing threads to obtain a first object list. According to the method, the GB-level data geographical screening efficiency and cost are improved, geographical screening and text filtering decoupling are achieved, and the LBS judgment speed is increased.

The application discloses an LBS parallel judgment method, system, device and storage medium, comprising obtaining and performing parallel matching on user location information and service text, and performing intersection processing on a first object list and a second object list obtained through matching to obtain a target object list; matching the user location information comprises: obtaining a first data set comprising object identification and location identification from a remote cache; starting multiple processing threads and distributing the first data set to the multiple processing threads after sharding; each processing thread reads geographic range information from a near cache composed of off-heap memory and a shared disk according to the object identification and the location identification, and matches the user location information with the geographic range information to determine a matching object; and the matching objects of the multiple processing threads are summarized to obtain the first object list. The application improves the efficiency of GB-level data geographic screening and takes into account the cost, realizes decoupling of geographic screening and text filtering, and improves the LBS judgment speed.

The application discloses a scoring prediction method based on end-to-end text filtering, comprising: obtaining a to-be-predicted review text, determining a user and a commodity corresponding to the review text; obtaining at least one historical review text of the user and at least one historical review text of the commodity; obtaining a word group level convolution feature vector corresponding to the review text and a word group level convolution feature vector corresponding to the historical review text; obtaining a feature vector corresponding to the review text and a feature vector corresponding to the historical review text; determining the similarity between the review text and the historical review text according to the feature vectors; filtering the historical review text based on the similarity, and calculating a similarity text feature vector according to the filtered historical review text; fusing the feature vector of the review text and the similarity text feature vector to obtain a global text feature vector; and determining the predicted score of the review text according to the global text feature vector. The application can improve the scoring prediction performance and precision of the review text.

The invention discloses a PDF (Portable Document Format) file cross-software sequence processing method and device and a related medium, and the method comprises the following steps: carrying out structured analysis processing on a PDF file to be processed, and then carrying out non-text filtering processing to obtain a text object data set; performing logic structure reconstruction by utilizing the text object data set to obtain a paragraph structure data set; performing intermediate format serialization processing and deserialization processing on the paragraph structure data set to obtain a target paragraph data set; respectively performing consistency marking and reversibility control processing to obtain a version control data set; and performing incremental updating on the version control data set to generate a PDF file target processing result. According to the method, consistency marking and reversibility control processing are carried out on the target paragraph data set obtained through calculation to obtain the version control data set, and then incremental updating is carried out on the version control data set, so that the consistency of a logic structure and a typesetting style can be kept when the same PDF file is transmitted between different software.

The invention discloses an alarm text filtering method and device, equipment and a storage medium. The method comprises the following steps: firstly, carrying out serialization processing on an alarm text by adopting a bidirectional coding layer, and converting the text into a sequence containing a plurality of word vectors; then, BiLSTM is used to encode the word vector sequence to generate a hidden state representation containing a plurality of hidden vectors. Thirdly, performing linear transformation on each hidden vector through a full connection layer to obtain multi-label prediction corresponding to different entity type label spaces and scores of the multi-label prediction; and then, by means of a CRF layer and a transfer matrix, carrying out joint decoding on all label prediction results, capturing a dependency relationship among labels, and selecting a label sequence with the highest conditional probability as a final target label sequence. And finally, matching and filtering the alarm text by utilizing a generalized regular expression rule corresponding to the target label sequence. According to the invention, missed judgment of important alarms or misjudgment of irrelevant information can be reduced.

The present application relates to the technical field of municipal drainage pipe network, and particularly relates to a drainage system CAD-GIS data conversion and standard marking method, which comprises the following steps: analyzing CAD data and extracting, extracting data in CAD according to types, and storing text files; processing CAD data, generating directed graph structure data from the extracted CAD data, generating KDTree search tree structure data from the X and Y positions of the text marking, filtering and finding the text near the midpoint of the preset inspection well or pipeline to obtain correct text marking content; converting the elevation text into real elevation, and obtaining the start point and end point elevation data of the preset inspection well, using the directed graph data structure and the extracted elevation data of the preset inspection well to analyze and sort the pipeline elevation and pipe diameter data; and converting into shp data. The method can greatly improve the information input efficiency of the drawing of the inspection well and the pipeline, and standardize the marking.

The invention discloses an aspect-level sentiment classification method based on context text generation and filtering. The invention provides an aspect-level sentiment classification method based on context text generation and filtering in order to solve the problems of few samples and low resources in part of fields in aspect-level sentiment classification tasks. The method comprises the following steps: firstly, carrying out context text generation by virtue of a Conditional Bert language model; secondly, a text screening and filtering algorithm is used for screening and filtering context text generation data to generate a high-quality text, and a data set is expanded; and finally, combining context text generation and filtering algorithm data to enhance the generation model. Experimental results show that the method provided by the invention effectively expands the data set, and improves the effect of aspect-level sentiment classification.