51 results about "Text retrieval" patented technology

The present disclosure provides a text recall method, a training method, device and equipment of a text recall model, relates to the technical field of computers, and particularly relates to the technical fields of artificial intelligence, large models, natural languages, deep learning and the like. The specific implementation scheme is as follows: a target query instruction is acquired; the target query instruction is input into a trained intent recognition model to obtain a first target vector output by a first target network layer in the intent recognition model; the target query instruction is input into a main model of a text recall model to obtain a prediction result output by the main model; the prediction result is used to indicate a prediction probability of a word unit in a word table library at a predicted word unit position, the main model comprises a second target network layer, and an input vector of a next network layer of the second target network layer is determined based on a second target vector output by the second target network layer and the first target vector; and a target recall text is generated according to the prediction probability of the word unit in the word table library at the predicted word unit position.

This application provides a method, apparatus, device, medium, and program product for generating training data, which may involve artificial intelligence technology and computer technology. The method includes: inputting a first document from a data source corresponding to a vertical domain text retrieval model into a first large language model to obtain a query statement corresponding to the first document; querying the query statement in the data source to retrieve N second documents; where N is an integer greater than 1; selecting M target second documents from the N second documents that constitute negative samples with the query statement; where M is a positive integer; and for each of the M target second documents, constructing a triplet training data with a partial order relation for a general text retrieval model using the query statement, the first document, and the target second document. This automatic generation method of training data can improve the efficiency and accuracy of training data generation compared to manual methods.

The application relates to the technical field of knowledge graphs, and specifically discloses a fabric cross-modal image-text retrieval method based on a knowledge graph and a storage medium, which comprises the following steps: constructing a textile fabric structured knowledge graph; performing semantic modeling on the textile fabric structured knowledge graph according to a TransR model; performing semantic structure extraction on the textile fabric structured semantic model according to a graph attention network; extracting deep semantic features of a textile fabric image; constructing an image-entity bidirectional contrast loss function according to a textile fabric entity semantic structure and the deep semantic features of the image; constructing a multi-task joint loss function according to the bidirectional contrast loss function and the structured semantic model; and training the spatial alignment matching degree of the entity semantic structure and the deep semantic features of the image according to the multi-task joint loss function to obtain a fabric cross-modal image-text retrieval result. The fabric cross-modal image-text retrieval method based on the knowledge graph can be applied to the cross-modal image-text retrieval of fabrics.

This invention belongs to the field of text retrieval technology, specifically referring to an automatic retrieval system for digital economy texts. The system includes a basic task instance identification module, a key instance association mining module, a tag local score estimation module, a global tag configuration module, a noise tag configuration generation module, and a precise digital economy text retrieval application module. This solution uses two dedicated multilayer perceptrons to calculate semantic and syntactic dependency weights and fuse them to obtain the final association strength, achieving precise capture of digital economy instances and their associations. It combines global tag configuration to quantify the tightness of instance tag associations, and uses the noise tag configuration generation module to transform the global optimal tag configuration problem into a binary classification problem. By constructing a contrastive loss function and optimizing it through backpropagation, it efficiently solves the problem of computational explosion, achieving concept-based precise retrieval and intelligent decision support, providing users with an efficient digital economy information acquisition experience.

This invention relates to the field of intelligent fabric matching and recognition technology, and discloses a fabric matching method and system based on multimodal information matching. The method includes: semantically structuring and enhancing user input text; acquiring the original fabric image and performing mask extraction, texture enhancement, and normalization processing; calculating the matching degree based on a cross-modal semantic alignment mechanism, and outputting fabric recommendation results by combining multi-factor joint ranking. Compared to existing technologies that use simple text retrieval or single-channel feature matching, especially when user descriptions are non-standard, fabric image textures are complex, or structures are repetitive, failing to achieve accurate fabric location and consistent semantic matching between text and images, this application improves fabric retrieval accuracy in open descriptive contexts by introducing an industry ontology-driven semantic enhancement mechanism and a saliency-guided cross-modal matching architecture, achieving multi-granular modeling and fine-grained semantic alignment of fabric semantic features.

This invention belongs to the field of information retrieval technology, and specifically discloses a vertical domain image retrieval method based on text and semantic fusion. This solution uses the image filename, storage path, and visual description associated with each image as multi-source text information, and constructs an inverted index corresponding to the images. This effectively utilizes the implicit classification and semantics within folder hierarchies, improving resource utilization. Furthermore, in a hybrid mode of text retrieval and deep semantic retrieval, it significantly improves retrieval accuracy and efficiency, especially for images with non-standard naming, resolving the fundamental contradiction of existing retrieval methods that struggle to balance efficiency and accuracy. In addition, by assigning weight coefficients to different sources within the multi-source text information, this solution consistently ensures that images matching the query terms and image filenames receive the highest scores and are ranked at the top, significantly improving query accuracy and efficiency while preserving the user's original query habits.

This application relates to a model data processing method, apparatus, computer device, computer-readable storage medium, and computer program product. The method includes: acquiring first contrastive learning sample data consisting of a query text, a list of positive sample document texts, and a list of negative sample document texts; performing sample generation processing based on the query text to construct second contrastive learning sample data consisting of the query text, a list of similar query texts, and a list of dissimilar query texts; training an initial text vector retrieval model based on the first contrastive learning sample data to obtain a first text retrieval model; and fine-tuning the first text retrieval model based on the first and second contrastive learning sample data to obtain a second text retrieval model. The second text retrieval model in this application can better understand and capture the similarities and differences between different query texts, thereby improving the robustness of the second text retrieval model to query texts.

The present application relates to the technical field of remote sensing image processing, in particular to a kind of explicit and implicit semantic collaborative guidance's cross-modal remote sensing image-text retrieval method, comprising: text is masked with context-aware TF-IDF weighting strategy processing;Explicit branch will masked text as implicit semantic embedding and image are fine-grained alignment;Implicit branch by the implicit semantic embedding produced during pre-training and image are adaptively fused features, at the end of implicit branch, integrate a lightweight differentiable adaptive binary projector;Contrast loss produced by explicit branch provides alignment supervision signal to implicit branch to perform final retrieval matching;The present application effectively combines the strong generalization ability of implicit alignment and high precision of explicit alignment through the collaborative interaction of two branches, realizes the high-precision, high-efficiency retrieval of remote sensing image and text, and significantly improves the retrieval performance in complex remote sensing scenarios.

The application discloses a target text retrieval method and device. The method comprises the following steps: acquiring a searched text set and a to-be-searched text, inputting the searched text set and the to-be-searched text into a preset model for analysis, obtaining a first feature vector set corresponding to the searched text set and a second feature vector corresponding to the to-be-searched text, wherein the preset model is obtained by training a pre-acquired positive and negative sample data set, and the to-be-searched text is associated with the text in the searched text set; determining a first feature vector closest to the second feature vector in the inner product distance from the first feature vector set; and determining the text corresponding to the first feature vector as a searched target text. The application solves the technical problem of high model training cost caused by the need of a large amount of artificial label data for training data.

The application relates to the field of artificial intelligence and knowledge graph, and discloses a time sequence event question and answer method based on structured subgraph retrieval, which comprises the following steps: performing text retrieval on event facts to obtain background information according to a question and answer request; analyzing the request in combination with the background information, extracting corresponding event structure constraints and time constraints; strictly screening event facts from a time sequence knowledge graph according to the constraints, and constructing an event subgraph with time attributes; performing adaptive retrieval generation on the subgraph, uniformly compressing along the time dimension or supplementing knowledge by using text retrieval, and obtaining a final subgraph after adaptive processing; and generating a time sequence question and answer result based on the subgraph. The application retains the structure and time constraints of the time sequence graph, overcomes the input length limitation of the model, and can significantly improve the question and answer accuracy and stability without fine-tuning the model.

This invention discloses a method and system for generating auxiliary diagnostic results and reference prescriptions. First, a medical record database and a treatment strategy database are constructed and stored in a vector database and a text retrieval database, respectively. For a target medical record, vectorization and textualization are performed. Then, candidate medical records and treatment strategies are retrieved from the medical record vector database, medical record text database, treatment strategy vector database, and treatment strategy text database via four channels. Subsequently, the RRF algorithm is used to fuse and sort the vector and text channels, and the candidate results are further refined based on a pre-trained semantic ranking model to obtain a small number of reference medical records and treatment strategies highly relevant to the current condition. Finally, prompt words are constructed by combining the target medical record, reference medical records, and treatment strategies to generate candidate results such as TCM diagnosis, TCM syndrome type, Western medicine diagnosis, and TCM prescriptions, serving as auxiliary references for doctors' clinical decision-making. This significantly improves the accuracy, interpretability, and efficiency of intelligent assisted diagnosis and treatment.

The application discloses a kind of based on thick and thin granularity modal interaction's picture-text retrieval method and system, comprising: extracting image global feature to sample image, extracting the word fine-grained feature of each word and text context feature to sample text;According to the similarity between image global feature and text context feature, obtain global cross-modal similarity;According to image global feature and the word fine-grained feature of each word, obtain local cross-modal similarity;Design multi-scale modal contrast loss function, according to global cross-modal similarity and local cross-modal similarity respectively obtain multi-scale modal contrast loss function under coarse granularity and fine granularity, train retrieval model;Image data or text data to be searched is obtained using the trained retrieval model to obtain the retrieval result, improve retrieval speed, efficiency and retrieval quality.

The application discloses a large language model training method and device based on range position coding and a medium, and relates to the technical field of artificial intelligence and natural language processing; the method comprises the following steps: constructing a base model network architecture, removing explicit position coding; configuring a range attention mechanism, assigning an exponentially distributed review range to different attention heads in a multi-head self-attention layer; constructing a sparse attention mask matrix according to the review range, and forcing the model to only pay attention to historical information within the review range; through the structured sparse design, the sequential perception of the full sequence is realized by using the recursive long-short context comparison mechanism, the model is endowed with the original length extrapolation capability and the high-precision long text retrieval capability while the calculation amount is greatly reduced, and the calculation redundancy and numerical instability problems of the traditional explicit position coding in the ultra-long context scene are solved.

This invention discloses a method for accelerating AI agent tool invocation based on a two-level caching mechanism, relating to the field of data processing technology. First, the tool set is vectorized to generate dense semantic vectors and BM25 sparse representations, constructing a tool cache database. Upon receiving a user query, a dual-path retrieval is performed in the historical question-answer pair cache database using semantic retrieval and BM25 text retrieval. If the matching degree exceeds a threshold, the historical answer is directly returned; otherwise, a dual-path retrieval is performed in the tool cache database to obtain candidate tools. After a large language model selects the tool for execution and generates the final answer, the new question-answer pair is written to the historical question-answer pair cache, achieving a closed-loop update. This method is suitable for efficient AI agent invocation in multi-tool scenarios.

The present disclosure provides a text retrieval model training method, a text retrieval method and related devices, and relates to the technical field of artificial intelligence. The method comprises: compressing each pair of query word samples and candidate text samples into a string sample; processing the string sample by using a feature extraction sub-model in a text retrieval model to obtain hidden layer features of each word unit in the string sample; calculating feature weights of the hidden layer features of each word unit by using a weight determination sub-model in the text retrieval model; calculating a relevance score between the query word samples and the candidate text samples by using a similarity calculation sub-model in the text retrieval model based on the feature weights and the hidden layer features of the corresponding word units; and training the text retrieval model in a contrast learning manner based on the relevance score. The method can improve the training accuracy and precision of the text retrieval model.

The application provides a sports question and answer method and device based on knowledge retrieval agents, the method comprising: receiving a sports question input by a user; inputting the sports question into multiple knowledge retrieval agents, each knowledge retrieval agent retrieving information related to the sports question from a corresponding information source to obtain at least one retrieval result, wherein the multiple knowledge retrieval agents include at least two of the following: a text retrieval agent, a data retrieval agent, and an online retrieval agent, the information source corresponding to the text retrieval agent including a knowledge base constructed based on sports-related texts, the information source corresponding to the data retrieval agent including a relational database constructed based on related data of sports videos, and the information source corresponding to the online retrieval agent including the Internet; and generating an answer to the sports question based on the sports question and the at least one retrieval result.

The application relates to an interactive short text topic recognition method based on a DW-ICNN and a matter logic graph, and the method comprises the following steps: acquiring interactive short texts, and preprocessing the interactive short texts; setting a sliding window; in the sliding process of the sliding window, the preprocessed interactive short texts are combined, then the texts in the window are subjected to event matching, and based on the event matching result, all the interactive short texts are distributed to event bags; the feature marks of speakers are acquired, the feature marks and the text information in the event bags are input into a DW-ICNN model, and a recognition result is acquired; according to the recognition result, the short texts of the same event are merged to form virtual long texts, relevant information of the long texts is searched, the long texts and the relevant information are input into a QWEN2.5 large model, an event and an evolution relationship are acquired, and a matter logic graph is constructed according to the event and the evolution relationship. The application realizes accurate perception of community topics and demands.

The invention provides a medical document retrieval enhancement generation method and system based on a father-child node structure, and relates to the technical field of medical document retrieval, and the method comprises the following steps: analyzing an input medical document to obtain a medical text unit sequence; constructing three layers of medical semantic tree nodes; performing quality inspection on each paragraph node to obtain a retrieval abstract; storing the retrieval abstract into a vector database, and storing atomic proposition tetrad contents and paragraph keywords into a full-text retrieval library; routing user query to a vector database to execute semantic retrieval, and simultaneously executing keyword retrieval in parallel in a full-text retrieval library to obtain a primary retrieval result set; and calculating relevance scores of the primary retrieval result set and the user query by adopting a reordering model, and performing descending order arrangement according to the scores to obtain a final retrieval result list. The problems that an existing RAG system is insufficient in document structure perception in a medical document processing scene, the integrity of medical facts is lack of reliable guarantee, and the multi-granularity retrieval cooperation capability is insufficient can be solved.

This application provides a text retrieval method and apparatus. The text retrieval method includes: acquiring a query question text; determining at least one target question text matching the query question text from various reference question texts, and determining a first knowledge document corresponding to each target question text, wherein the reference question texts are questions corresponding to pre-set knowledge documents and stored in a pre-built target knowledge base; determining at least one second knowledge document matching the query question text from various knowledge documents in the target knowledge base; and determining a target knowledge document corresponding to the query question text based on each first knowledge document and each second knowledge document. When retrieving target knowledge documents similar to the query question text, not only the knowledge document itself can be referenced, but also the reference question texts corresponding to the knowledge document, thereby improving the relevance between the retrieved target knowledge documents and the query question text.

This invention discloses a hierarchical retrieval method, storage medium, and computer device, belonging to the field of document processing technology. It includes: S100, obtaining the hierarchical paragraph parsing results of the input document; S200, retrieving the title paragraphs in the parsing results according to the user's query, generating a set of title paragraphs based on the retrieval results, determining whether the title paragraph set is empty, and if not, outputting a semantically complete paragraph; if so, proceeding to step S300; S300, retrieving the body text paragraphs in the parsing results according to the user's query, and outputting a semantically complete paragraph or summary based on the retrieval results. This invention, by combining the hierarchical parsing results of the document, can improve the completeness of text retrieval results in large-scale model retrieval and enhance document question answering, thereby improving the correctness and completeness of the generated answers in document question answering.

The application belongs to the technical field of image-text retrieval, and discloses a multi-view contrast learning fine-grained image-text retrieval method based on CLIP and related equipment. The visual view and the text view are enhanced, a visual encoder and a text encoder based on residual attention are combined to extract features, the defects of insufficient capture of local high-frequency features of a traditional ViT encoder are made up, feature collapse is avoided, contrast pairs are generated based on a contrast strategy and total loss is calculated, encoder parameters are synchronously updated through back propagation, the correlation within the mode can be fully tapped, the negative influence of network noise image-text pairs is weakened, cross-modal semantic correlation is accurately established, better cross-modal alignment is realized, and unified and robust user representation is learned from massive unlabeled multi-modal data.

The application discloses a text retrieval method, system, device and storage medium. It relates to the technical field of artificial intelligence, and the method comprises the following steps: receiving a retrieval request of a user and performing vectorization processing to obtain a retrieval vector; performing retrieval in a preset knowledge index based on the retrieval vector and determining a target document; constructing and generating prompt information based on the target document and the retrieval request; inputting the generated prompt information into a large language model for retrieval enhancement generation processing to obtain a retrieval result. The application solves the problems that the existing retrieval result is scattered, it is difficult to match deep intent, and the result credibility is insufficient when processing long documents. The overall target document is retrieved and determined in the knowledge index, and the complete information unit is locked. Then, the target document and the retrieval are combined to construct and generate a prompt, which guides the large language model to output an accurate and complete answer. Finally, the result with clear document traceability is output, and the accuracy, integrity and credibility of the retrieval are improved.

This application relates to a knowledge graph-based retrieval method and related apparatus, comprising: acquiring raw text data; extracting entities from the raw text data and identifying relationships between entities using a first language model to obtain triplet data and descriptive text data corresponding to each entity; constructing a knowledge graph and target vector data; performing similarity retrieval in the knowledge graph and target vector data based on first text information input by the user to obtain at least one second text information; and inputting the second text information into a second language model so that the second language model outputs text retrieval results based on the second text information. The solution provided in this application, by automatically extracting entity relationships to construct a knowledge graph and combining it with vector data for retrieval, can improve the efficiency of knowledge graph construction and retrieval accuracy, reduce model illusions, and improve the accuracy of text retrieval results.

The application provides a kind of image retrieval method and device based on multimodal feature fusion, the method comprises: extracting the global feature and local feature of retrieval text, retrieval sketch and candidate image respectively;The global feature of the retrieval text and the retrieval sketch is fused, and the global similarity between the fused feature and the global feature of the candidate image is calculated;Based on the local feature of retrieval sketch and candidate image, the corresponding target area feature is determined respectively;The local feature of retrieval text and the target area feature of retrieval sketch are interactively fused, and the fusion result is matched with the target area feature of candidate image, to obtain local similarity;According to global similarity and local similarity, determine the retrieval result from candidate image, not only give full play to the complementary advantage of text semantic and sketch visual information, but also improve the accuracy and robustness of image retrieval.

The application discloses a structure perception UAV image-text retrieval method based on Riemann guiding alignment, which comprises feature extraction, structure representation construction, cross-modal alignment model building and joint loss function design, and is composed of a semantic branch and a structure branch: the semantic branch is used for acquiring global semantic embedding of images and texts; the structure branch generates a symmetric positive definite matrix through covariance modeling, describes spatial dependence and geometric relationship in a UAV scene, and is mapped to a tangent space under a logarithmic Euclidean metric, so that structure-preserving feature representation is realized; a marginal distribution and conditional distribution alignment mechanism is further constructed, and a semantic structure fusion strategy is combined to adaptively reduce the distance of cross-modal features in a unified embedding space. The application can relieve geometric distortion and modal mismatch problems caused by Euclidean modeling, enhance structure expression and semantic consistency in a complex scene, and thus improve the accuracy and robustness of UAV image-text retrieval.

The application provides a kind of financial and tax data intelligent dialogue and query system based on large language model, it is related to financial and tax data processing and natural language processing technical field, first, the intermediate adaptation layer connecting large language model and financial and tax business database is built, and the database information is converted into the form that large language model can recognize.The large language model is trained in the field of finance and tax, so that it can process user financial and tax problems and generate query instructions.Through the intermediate adaptation layer, the data query is executed, and the original numerical result is obtained.The model is called to analyze the result and generate a preliminary interpretation, and the final result is formed after correction combined with the financial and tax rule text retrieval library and pushed to the user.The application improves the efficiency and accuracy of financial and tax data query, and provides comprehensive and professional financial and tax information for users.

The application relates to a data catalog retrieval method, an electronic device and a computer readable storage medium. The method is applied to a data catalog platform, and the data catalog platform pre-stores a metadata database, a text library, a vector library and a key-value library. The key-value library stores key-value pair data with a service name as a key and service information other than the service name as a value. The method comprises the following steps: receiving original retrieval information input by a user; calling a large model to perform semantic understanding, keyword extraction and term normalization processing on the original retrieval information, and generating standardized retrieval information; performing field matching retrieval in the metadata database according to the standardized retrieval information; performing full-text retrieval in the text library and vector similarity retrieval in the vector library according to the original retrieval information and the standardized retrieval information; matching the standardized retrieval information with the keys in the key-value library and performing relevance calculation on the values corresponding to the candidate keys; and performing fusion processing on the multi-path retrieval results and outputting candidate service information.

The application discloses a text chunking method and device, a storage medium and a computer device. The method comprises the following steps: obtaining an initial text, and performing segmentation processing on the initial text to obtain a plurality of target texts; performing division processing on the target texts by using identifiers to obtain a plurality of initial sentences; performing splicing processing on the plurality of initial sentences to obtain a plurality of target sentences, and determining a semantic threshold of the target sentences based on the plurality of target sentences, wherein the semantic threshold is used for representing the semantic similarity of different target sentences; and performing chunking processing on the target texts according to a text chunking strategy based on the semantic threshold to obtain a plurality of target chunks, wherein the text chunking strategy is used for representing the chunking rule of the target texts, so that the technical problem of low text retrieval accuracy is solved, and the technical effect of improving the text retrieval accuracy is achieved.

The present application relates to the technical field of text retrieval, in particular to a retrieval enhancement and constraint guidance method and system for high-consistency text generation. The present application generates text segments sentence by sentence through a streaming generation mechanism and captures semantic units of each segment in real time, can closely combine actual needs in the generation process, effectively realizes dynamic adjustment in the generation process, through the aid of preset constraint rules and the feedback mechanism of semantic units, the system can continuously optimize the reasoning logic and the generation boundary at each stage of generation, not only improves the quality of text generation, but also provides quality guarantee for the final output through consistency check, forms a closed-loop feedback mechanism, so that the generation process can respond to changes in information in real time, handle complex information relationships, and effectively fill in the required factual gaps.