376 results about "Headword" patented technology

The invention discloses a large model knowledge base construction method based on multi-granularity retrieval. The method comprises the steps of obtaining a target text and analyzing a document structure, splitting and reorganizing text data according to three levels of titles, fragments and sentences, and constructing three levels of knowledge of the titles, the fragments and the sentences; using a large language model to construct a generative index as question-level knowledge according to the sentence-level knowledge; using a vector model to generate a first vector representing index keys of fragment-level knowledge and / or sentence-level knowledge, and a second vector representing index keys of title-level knowledge and / or question-level knowledge; and respectively storing the first vector and the second vector into a vector database, and storing the index values and the index types corresponding to the first vector and the second vector into a relational database. According to the method, multi-granularity document knowledge base construction is achieved, the expression modes of knowledge under different granularities are expanded, the accuracy of user query and knowledge matching is improved, and the problem that information is lost or incomplete in an existing method is solved.

The invention discloses a science and technology public text intelligent classification and service method and device based on deep learning. The method comprises the steps that multi-source science and technology public text data are acquired and preprocessed; extracting keywords by adopting a keyword extraction algorithm, and splicing the keywords with the public text title to form enhanced text features; constructing a multi-dimensional public text classification system and performing data annotation; performing feature extraction and fine adjustment by adopting a BERT pre-training model to obtain a classification model; automatically classifying the newly-added public texts and visually presenting the newly-added public texts; and generating a personalized recommendation result based on the user portrait and the public text feature index. The invention further relates to a technical scheme of multi-objective quality diversity optimization, heterogeneous resource allocation and fusion of the LPLC2 neural network and the BERT. The technical problems that a traditional method is limited in complex semantic understanding ability, single in classification dimension and lack of an integrated solution are solved, and the accuracy of science and technology public text classification and the intelligent level of service are improved.

The invention provides a semantic enhancement adaptive partitioning method and system for natural resource large model questions and answers, and aims to solve the problems of difficulty in term boundary recognition, damage to semantic integrity and the like. According to the method, three core technologies including theme perception coarse-grained paragraph division, self-adaptive sliding window theme hierarchy division and embedded perception context self-adaptive text segmentation are fused. The system firstly analyzes a natural resource long text structure, identifies titles and theme levels and aligns associated contents; paragraphs are extracted according to a theme perception strategy and are subdivided into sentence sets according to grammar rules; an improved sliding window mechanism is adopted to divide sentences into window sentence block groups. The method is characterized in that a dynamic aggregation threshold mechanism is introduced, the semantic association degree between adjacent sentence blocks is calculated through an embedded perception context semantic segmentation technology, whether the sentence blocks are combined or not is judged by combining a similarity distribution change trend and a dynamic adjustment threshold, self-adaptive delimitation of semantic boundaries is achieved, and text blocks which are clear in structure and coherent in semantics are generated.

The invention provides a document segmentation method and device and electronic equipment, relates to the technical field of artificial intelligence such as natural language processing, large language models, intelligent search and knowledge maps, and can be applied to scenes such as knowledge base construction, intelligent question and answer systems and document management systems. According to the specific implementation scheme, document analysis is conducted on a to-be-processed document, an analysis result of the to-be-processed document is obtained, and the analysis result comprises layout information and hierarchy information of a title; a title tree is constructed based on the hierarchical information of the titles, and the titles associated with nodes in the title tree are multi-level composite titles; and based on the layout information and the title tree, splitting and recombining the document region of the to-be-processed document according to a given maximum slice length to obtain document slices of the to-be-processed document. According to the method and the device, the document slicing precision and flexibility can be remarkably improved, and the document readability and the user experience are enhanced.

The invention discloses a short text clustering method and device based on a large language model, and belongs to the technical field of natural language processing, and the method comprises the steps: converting a short text into vector representation, and carrying out the preliminary clustering of the vector representation through a clustering algorithm, and obtaining an initial category; for all the corresponding short texts in each initial category, generating an abstract by using a large language model, generating a plurality of theme titles for summarizing the core meanings of the texts according to the abstract, respectively taking the theme titles as sub-titles, inputting the plurality of sub-titles into the large language model, and carrying out theme combination to obtain a parent title, constructing the parent title and the child titles into a hierarchical theme system; the large language model allocates to-be-classified short texts to one or more categories according to the hierarchical topic system to complete short text clustering; and performing quality evaluation and error pattern analysis on a short text clustering result to optimize a clustering algorithm and a large language model. According to the method, the theme of the short text can be automatically found and classified, and the classification accuracy and efficiency are improved.

The large model knowledge base retrieval method based on multi-granularity retrieval comprises the steps that a knowledge base based on three levels of knowledge of titles, fragments and sentences is constructed, and the knowledge base comprises a vector database storing first vectors and second vectors and a relational database storing index values and index types corresponding to the first vectors and the second vectors; generating a vector value queried by a user by using a vector model, respectively querying a plurality of query index vectors with the highest similarity with the query vector value in the first vector and the second vector, and obtaining query index values and query index types corresponding to the query index vectors from the relational database as candidate retrieval results; and jointly constructing a query prompt word by using a plurality of candidate retrieval results and user query, and inputting the query prompt word into the large model to obtain a final query reply. According to the method, the precision, the efficiency and the intelligent level of a retrieval system are remarkably improved, and the problems of low precision, poor query adaptability, slow knowledge base updating and the like in the prior art are solved.

The invention belongs to the technical field of document content extraction, and particularly discloses a document content extraction method and system based on multimodal model collaboration, a terminal and a medium. Comprising the following steps: identifying the type of an input to-be-processed document, and judging the document type; on the basis of the type identification result, calling a multi-modal model to analyze the document content, and outputting space coordinates, visual features and semantic features of document elements; generating a content sequence according with a reading habit through a semantic sequence reconstruction algorithm; paragraph boundary detection, paragraph recombination and semantic association modeling of charts and texts are completed based on the multilayer attention network and the graph neural network; grammar error correction, format optimization and title hierarchy generation are carried out by using a large language model and a hierarchical classification network; and converting the identification result into a structured output file. According to the method, the processing requirements of different types of documents can be considered, and high-precision analysis and efficient output are realized under the scenes of complex layouts, multiple languages and formula tables.

The invention provides a highway intelligent operation and maintenance question-answering system based on a large language model, and belongs to the technical field of natural language processing. The system takes a large language model as a core reasoning engine and combines a domain knowledge base and an RAG technology to realize accurate question and answer of highway operation and maintenance; the method comprises the following steps: based on original knowledge data cutting, generating a title through a large language model, and customizing a knowledge base; receiving query, analyzing an intention by using a large language model, and matching to generate a function; the query is rewritten by using a large language model, dense and sparse vector query is generated, and a double-layer retrieval mechanism is formed; a two-step recall mode is utilized, coarse-grained recall is firstly carried out, then a recall result is subjected to fine-grained optimization through a screening mechanism, and a reasoning text is generated; and finally, inputting the query and reasoning text into the large language model, and generating an optimal answer through single-round and multi-round questions and answers. According to the method, the professionality and reliability of answers are enhanced, and the technical problem that answers are incomplete and inaccurate in an existing question and answer system is solved.

The invention relates to a PICO extraction method and device based on a large language model and a storage medium, and is applied to the technical field of medical information, and the method comprises the steps: obtaining and preprocessing titles and abstracts of a large number of medical literatures, then combining preset cues related to PICO with texts to construct an instruction data set, and storing the instruction data set in a database; a text in an instruction data set is subjected to word segmentation processing to be converted into an input sequence, a large language model framework is constructed, a false positive suppression loss function is set, the false positive rate of PICO element extraction by the model is remarkably reduced, the large language model framework is trained through the input sequence, and a trained PICO element extraction large language model is obtained. Preprocessing medical text data to be retrieved, inputting the preprocessed medical text data into the trained model, and outputting a PICO extraction text by the model; and finally, each PICO classification result is output through the PICO classifier. The data are data related to the crowd / problem, intervention, comparison and result in the to-be-retrieved data.

The invention belongs to the technical field of text processing, and particularly relates to a method and system for retrieving DOCX document content based on keywords, which comprises the following steps: analyzing an Office Open XML structure of a DOCX document, combining with multi-dimensional features such as style names, and utilizing a title classification score model to accurately distinguish a title and a text, so that a semantic hierarchical structure of the document is effectively reserved; and secondly, a multi-level semantic extension mechanism is introduced, and a Sension-BERT, a HowNet knowledge base and a Word2Vec model are fused, so that intelligent extension of synonyms and synonyms of keywords is realized, and the recall rate and semantic understanding ability of retrieval are remarkably improved. And in addition, a BM25 model is combined with paragraph length normalization and structure position weight to calculate a correlation score, so that retrieval results are sorted more accurately and reasonably. The construction of the reverse index is combined with the position coding and compression optimization strategy, and the retrieval efficiency and the storage performance are both considered.

The invention discloses a vehicle knowledge question-answering method based on a GraphRAG method and computer equipment, and the method comprises the steps: building a document tree structure based on the title hierarchy of a vehicle knowledge document, distributing a text identifier, and adding metadata information; converting the stored text units into text unit vectors by adopting a vector embedding model, and establishing a text vector index; calculating a similarity score between the query vector and the text unit vector, and determining a traditional RAG path text unit based on the similarity score; identifying entities and entity relationships of the text units according to the entity types and the relationship types by adopting a large language model so as to generate a vehicle knowledge graph; performing multi-level community division on the vehicle knowledge graph by adopting a Leiton community detection algorithm, and generating a structured community abstract according to each level of community; determining a GraphRAG path community abstract by adopting a large language model; and fusing the community abstract of the GraphRAG path with the text unit of the traditional RAG path and generating an answer.

The invention discloses an intelligent writing method based on multi-agent cooperation, and the method comprises the following steps: receiving documents of various formats uploaded by a user, extracting text and image resources, and understanding an image through a multi-modal large language model to generate a description text; constructing a cue word template by using a large language model, processing three types of input files, extracting key information, and constructing a knowledge model; receiving user input construction enhanced cue words, and calling a large language model to generate an article outline; chapter contents are generated one by one according to a title segmentation outline, and a writing execution agent generates a text according to requirements and automatically illustrates; and the MarkDown format output is converted into Word with a basic format. The method has the beneficial effects that intelligent writing is completed through multi-step cooperation, information is extracted from multiple sources, a model is constructed, an outline and content are efficiently generated, a diagram is automatically illustrated, finally, the format is converted, and convenient, efficient and functional intelligent writing services are provided for users.

The invention relates to a structured text analysis method and system based on title recognition and hierarchical abstracts, and the method comprises the steps: receiving a text, and obtaining sample data of a training language model; the vector database can perform vector dimension adjustment on the data of the stored text. The vector database takes vectors as basic storage units, and converts unstructured data into high-dimensional vectors through an embedding technology. And calling the data storage type information of the title text data set to determine the data storage type of the title text data set matched with the text storage mode. And obtaining a structure type of sample data of each text by utilizing a semantic segmentation script. And segmenting the document according to the identified chapter titles, and subdividing each chapter into a plurality of logic paragraphs to form structured text block units. The multi-level abstract generation model matched with the target model can fully mine level information and semantic association in the document. Therefore, the effect of analyzing the structured text by the intelligent question-answering system is improved.

The invention discloses a document segmentation method for large model retrieval enhancement generation, and relates to the technical field of document segmentation, the method comprises the following steps: taking any secondary title and text content under the secondary title as a to-be-analyzed object, and counting the number of characters in each to-be-analyzed object, when the number of the characters exceeds the preset number of the characters, paragraph splitting is carried out on the text content; and carrying out segmentation confidence calculation on each split part, and carrying out merging or segmentation processing on the adjacent parts based on the segmentation confidence corresponding to the adjacent parts. According to the method, the segmentation confidence coefficient of the adjacent parts is calculated, the document can be accurately and partially segmented, the segmentation effect is effectively guaranteed, and in addition, based on the segmentation confidence coefficient, the semantic integrity after segmentation and before and after segmentation can be effectively guaranteed.

The invention provides a document information extraction method and device, electronic equipment and a medium, and the specific implementation scheme comprises the steps: obtaining a to-be-processed document, the to-be-processed document being a document which does not contain a title identifier, and the title identifier being used for identifying the title level of each chapter title in the to-be-processed document; analyzing the to-be-processed document to determine text content information and text position information respectively corresponding to a plurality of chapter titles in the to-be-processed document; determining a title level corresponding to each chapter title according to the text position information corresponding to each chapter title; and calling the large language model to perform information extraction on chapter contents among the chapter titles in the to-be-processed document based on the text content information, the text position information and the title levels corresponding to the chapter titles to obtain an information extraction result.

The invention provides a long report generation method based on a large model, and belongs to the technical field of natural language processing and artificial intelligence, and the method specifically comprises the following steps: analyzing a report demand input by a user, extracting key information, and generating task description; automatically generating a report framework based on the large model and the domain knowledge base, wherein the report framework comprises chapter division, title generation and content summary; calling a large model to generate detailed contents in chapters, and introducing a context memory mechanism to ensure the continuity between the chapters; grammar check, logic check and style optimization are performed on the generated content, and the professionality and accuracy of the report are improved in combination with a domain term library; and finally, outputting the report content according to a format required by a user, and automatically generating auxiliary contents such as a directory, a chart and reference literature. The method has the advantages of high efficiency, high quality, flexibility, expandability and the like, and can be widely applied to long report generation tasks in the fields of medical treatment, law, finance and the like.

The embodiment of the invention provides a retrieval enhancement generation method and system based on multivariate fusion, and the method comprises the steps: firstly structuring a knowledge base document, generating a knowledge graph, and importing an ES to establish an index; inputting question sentences, carrying out word segmentation and other processing, querying instance nodes from the map, and outputting meeting conditions according to answer sentence patterns; if not, fragmenting and blocking the document according to a title level, obtaining candidate results through vector, ES and atlas retrieval, normalizing scores, merging and optimizing the scores of the candidate retrieval results according to a retrieval source, calculating comprehensive scores, and outputting the comprehensive scores in a descending order; and finally, carrying out semantic integrity aggregation on the combined candidate results, and inputting into a large language model to obtain a final answer. According to the method, the relevance of the retrieved content is greatly improved, the semantic integrity of the retrieved content is improved, the model magic view problem is reduced, and the question and answer accuracy and the answer quality of a knowledge base are improved.

The invention provides a method and system for optimizing a commodity title and computing equipment, and the method comprises the steps: obtaining commodity information which comprises an original title of a commodity and a commodity attribute; obtaining related flow words associated with the commodity by using an algorithm recommendation system; based on the commodity information, first candidate traffic words used for being added into the original title are selected from the related traffic words through a large language model; based on the commodity information and the first recommended traffic word, utilizing a large language model to generate a second candidate traffic word used for being added into the original title; based on the commodity information, the first candidate traffic word and the second candidate traffic word, determining excluded risk traffic words by using a large language model to obtain a third candidate traffic word; and selecting an optimized traffic word from the third candidate traffic words by utilizing a large language model based on the commodity information, and adding the optimized traffic word into the original title, thereby generating a new title. According to the technical scheme, merchants can be better helped to screen drainage keywords, commodity titles can be more conveniently optimized, and commodity competitiveness is improved.

A computing system for monitoring language model compliance with a rubric of one or more output characteristics. The computing system includes processing circuitry configured to interface with a trained generative language model that receives input of a prompt including natural language text input and, in response, generates an output that includes natural language text output. The processing circuitry is further configured to monitor compliance of the generative language model with the rubric, by feeding the output of the generative language model to a rubric classifier configured to generate a predicted classification for an output characteristic in the rubric, and output the predicted classification.

The invention provides a method and system for automatically generating software codes, and the method comprises the steps: setting a menu table, a field table and an inter-table crossing relation, the menu table comprises column title content, module title content and module table name content, a plurality of column titles in the column title content are classified into one group and correspond to one module title, and the module title content corresponds to the module title; the module titles are in one-to-one correspondence with the module table names; the field table has user-defined field content and module table name content, and the user-defined field corresponds to a module table name; the inter-table crossing relationship is used for realizing crossing of the menu table and the field table; importing the menu table and the field table into an English-Chinese dictionary, performing statement decomposition, decomposing statements in the menu table and the field table into words and graphic icons, and realizing crossing of the menu table and the field table; and the data link model automatically generates codes according to the imported module menu and field paraphrasing. The invention also provides a system for realizing the method. The development efficiency can be improved, the error rate can be reduced, and the labor cost can be saved.

The invention provides a water conservancy knowledge structured extraction and verification method and device, and belongs to the technical field of artificial intelligence, and the method comprises the steps: carrying out the differential text processing of different formats of files, and generating an intermediate file; classifying the intermediate file into a regulation class or a non-regulation class based on a preset rule base; performing hierarchical title identification on the regulatory files to form entry knowledge blocks, and converting table contents into HTML (Hypertext Markup Language) knowledge blocks; performing semantic segmentation on the non-regulation file to generate knowledge blocks; performing knowledge block checking and filing, and marking an abnormal alarm block; converting the table knowledge blocks into natural language description by utilizing a large model; and positioning the context of the original text of the alarm knowledge block, and performing intelligent correction through a large model. According to the method, a traditional semantic analysis model and a large language model are creatively fused, a closed-loop process of preprocessing, extraction, verification and correction is formed, the problems of structured analysis and error correction of complex texts in the water conservancy field are solved, and the knowledge processing efficiency and accuracy are remarkably improved.

Embodiments of the present invention provide systems, methods, and computer storage media for identifying the relevant segments that effectively summarize the larger input video and / or form a rough cut, and assembling them into one or more smaller trimmed videos. For example, visual scenes and corresponding scene captions are extracted from the input video and associated with an extracted diarized and timestamped transcript to generate an augmented transcript. The augmented transcript is applied to a large language model to extract sentences that characterize a trimmed version of the input video (e.g., a natural language summary, a representation of identified sentences from the transcript). As such, corresponding video segments are identified (e.g., using similarity to match each sentence in a generated summary with a corresponding transcript sentence) and assembled into one or more trimmed videos. In some embodiments, the trimmed video is generated based on a user's query and / or desired length.

The invention relates to a similarity calculation-based scientific research topic selection recommendation method. The method comprises the steps of obtaining scientific and technical literature data and performing preprocessing; and extracting keywords from the abstract and the title by using a TextRank algorithm, and identifying technical keywords through experts. And carrying out vectorization on the keywords by using the trained word2vec algorithm. After keywords input by a user are obtained, vectorization is also carried out, and the cosine similarity between the keywords and each vector in the keyword vector table is calculated; and selecting the previous scientific and technical literatures and the corresponding scholars and technical keywords, and displaying the scientific and technical literatures and the corresponding scholars and technical keywords to the user through a visualization technology. According to the method, by calculating the cosine similarity between the keyword vectors, the theme in which the user is interested is determined, literatures, scholars, technologies and the like with high relevancy are recommended for the user, and the user is helped to quickly find high-quality literatures related to the theme in which the user is interested, leading figures in the field, possible research methods and the like.

The invention relates to the technical field of semantic understanding, in particular to an environmental impact report auxiliary auditing system based on semantic understanding, which comprises a text cleaning module, a chapter segmentation module, a rule generation module, a content auditing module and a result output module. According to the method, the paragraph structure and the layout identification are analyzed in a unified mode, key paragraph types in environmental influence chapters are efficiently recognized, the text integration accuracy is improved by combining inter-paragraph similarity calculation and repetition rate screening, chapter boundary accurate positioning and affiliation adjustment are achieved on the basis of format feature comparison of serial numbers and title styles, and the text integration efficiency is improved. A multi-layer parameter alignment rule set is constructed to support comprehensive verification of standard numbers, monitoring frequencies and periodic elements, an exception labeling mechanism is combined to complete difference item logic judgment and consistency evaluation, an auditing basis with rule driving and data comparison capabilities is formed, chapter positioning precision, element extraction comprehensiveness and logic exception recognition efficiency are improved, and the method is suitable for popularization and application. And the pertinence and the automatic processing depth in the auditing process are enhanced.

The invention relates to the technical field of data processing, in particular to a contract text key payment index extraction method based on natural language processing, which comprises the following steps of: performing structured chapter division on a contract text, constructing a chapter hierarchical tree, and extracting a key payment index of the contract text according to a matching relationship between multi-level path nodes of the chapter hierarchical tree and chapter title keywords, setting a candidate path containing payment information; extracting an action entity of each contract text in the candidate path, and extracting a correlation index corresponding to the action entity according to a condition clause and a condition type of the action entity; extracting adjacent words based on the relevance index of the action entity, and constructing a word relation graph by analyzing the part-of-speech features of the adjacent words and the current action entity; iterating the word relation graph according to a preset association rule, and correcting each node in the word relation graph; outputting a key payment index corresponding to the contract text according to the difference part of the word relation graph before and after correction; accuracy and efficiency of payment index extraction are realized.

A computer-implemented method for performing searches in a document database is disclosed. The method comprises automatically detecting a line of business associated with a user, receiving a text query from the user, and generating a query embedding from the text query. The method further comprises scoring entries in a reverse index using a hybrid scoring function. The reverse index comprises titles, title embeddings, sentences, sentence embeddings, and entity tags corresponding to documents in the document database. The hybrid scoring function is used to generate a score based both on a keyword match score between the text query and the reverse index and on a cosine similarity score calculated from embeddings in the text query and in the reverse index. The method also comprises ranking scores for sentences in the document database, and displaying a sentence associated with a top score to the user.

The invention relates to the technical field of building information, and provides a knowledge base construction and retrieval method and system based on a multi-source text in the building field, and the method comprises the following steps: a knowledge base construction stage: constructing a multi-dimensional metadata feature vector for a multivariate text based on a standard classification index table; the method comprises the following steps of: converting and segmenting a document, splicing an end clause with all superior title texts by utilizing a context inheritance algorithm to form a text unit with complete semantics, and performing dynamic filtering based on an analyzed query intention and a metadata vector at a user retrieval stage; then, in the screening set, performing fusion calculation on semantic vector similarity, keyword matching degree and authority offset weight based on effectiveness attribute and implementation time, and performing mixed retrieval and reordering on the text units; and finally selecting a text unit according to a sorting result and inputting the text unit into the large language model to generate answers. According to the method, high-precision and high-compliance intelligent retrieval and question answering of building domain knowledge are realized.

The invention discloses a context-separated false information detection method based on global information enhancement, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining an image and two titles to form a triple, inputting the triple into a multi-modal language reasoning model, inputting the triple with a neutral category index into a visual semantic reasoning model, and obtaining a visual semantic reasoning model; judging the triple with the contained category index as real information which is not separated from the context, and judging the triple with the contradictory category index as false information which is separated from the context; based on a visual semantic reasoning model, outputting relation scores # imgabs0 # and # imgabs1 # between the image and the two titles; the similarity # imgabs2 # between the two titles is calculated on the basis of a pre-training language model, and a triple with the # imgabs3 # and the # imgabs4 # both larger than a preset image-text threshold # imgabs5 # and the # imgabs6 # smaller than a preset text threshold # imgabs7 # is judged to be false information out of the context, wherein the # imgabs3 # and the # imgabs4 # are larger than the preset image-text threshold # imgabs5 # and the # imgabs6 # is smaller than the preset text threshold # imgabs7 #; according to the false information detection method, the false information out of the context can be better detected, and the detection accuracy is high.

The invention discloses a complex table question-answering method based on a tree structure, electronic equipment and a medium, and the method comprises the following steps: responding to header information of a complex table, and outputting a header tree structure by a large language model; decomposing the question into a plurality of keywords through a large language model, and aligning the keywords with the header tree structure to obtain a keyword-tree structure; and in response to the keyword-tree structure, guiding the large language model to carry out iterative reasoning through a React-Style prompt method to retrieve from the complex table, and outputting an answer. According to the method, the multi-level titles of the table are organized into the tree structure, so that the large language model can more clearly understand the hierarchical relationship of the table, and related information can be more accurately positioned when questions are answered.

The invention provides a form text boundary adaptive information extraction method and system in the technical field of computer information processing. The method comprises the following steps: S1, identifying a title frame from a form image; s2, calculating the aspect ratio of the title frame to determine the title arrangement direction of the table titles; s3, detecting a text detection box of the table image, and performing classification prediction of a text arrangement direction on the text detection box by combining a text direction model with a title arrangement direction; s4, controlling affine transformation of the table image through the text arrangement direction and the text detection box to obtain a corrected image; s5, identifying a table title in the corrected image, and matching a table type from the table type configuration file based on the table title; s6, extracting keywords from the corrected image based on the table type; and S7, processing the keyword to extract table information. The method has the advantages that the adaptability, accuracy and efficiency of table information extraction are greatly improved.