56 results about "Text searching" patented technology

The invention provides a full-text retrieval method and system fusing various types of documents, and relates to the technical field of information retrieval, and the method comprises the following steps: obtaining document representation through document content extraction and structure recognition, generating a cross-modal semantic vector by using word embedding and nonlinear transformation, constructing a hierarchical index and a cross-document association graph, and obtaining a full-text retrieval result; the basic correlation score is calculated after the query request is received, and the comprehensive score of the candidate content segments is calculated based on the association graph to determine the optimal retrieval result, so that unified representation and retrieval of heterogeneous documents are realized, the cross-document retrieval precision and relevance are improved, and the processing capability of a retrieval system on complex queries is enhanced.

A computer-implemented method for computational textual search to find and display search identified information in documents. Search queries are processed over one or more documents either at a server or user device using match schemes that produce both binary (match or no match) and non-binary (i.e. multiple matching values) results and have relative fuzziness relationships. A fuzzier match scheme implies more results and fewer false negatives and a less fuzzy match scheme implies fewer results and fewer false positives. This fuzziness relationship allows, without changing a search string, for users to quickly change from a match scheme to a fuzzier or less fuzzy match scheme depending on user evaluation of results as containing too many false negatives or too many false positives—without becoming a programmer of complex search string metadata or an expert user of advanced search capabilities.

The invention discloses a question answering method and device for credit review knowledge based on a large model and a medium, and the method comprises the steps: obtaining a to-be-processed credit review question, extracting a core keyword from the question through a natural language processing algorithm, and converting the question into a deep semantic vector through a semantic vector model; inputting the questions into an intention classification model, generating filtering labels, inputting the questions into a distributed full-text search engine, screening a target knowledge base according to the filtering labels, and performing retrieval in the target knowledge base based on the core keywords and the deep semantic vectors to obtain a preliminary retrieval result set; calculating the score of each preliminary retrieval result in the result set according to the core keyword, the deep semantic vector and the weight factor, and screening out final reference content in a descending order; and on the basis of credit and loan business compliance requirements, risk control rules and question and answer output specifications, the credit and loan prompt words are constructed, and the final reference content is input into the large model to obtain structured credit and loan review answers, so that the question and answer efficiency and accuracy are improved.

Systems or methods of the present disclosure may provide a design tool for adjusting designs implemented on programmable logic devices. The present disclosure includes receiving documentation and receiving a full text search. The documentation may include user guides, technical specifications, and design files such as HDL code, constraints, timing reports, and / or design assistant / rule violation (DRC) reports. The present disclosure also includes determining semantic search for embedding vectors based on the documentation and the full text search. Furthermore, the present disclosure includes providing the semantic search to a large language model (LLM).

The invention discloses a BEV-based scene-level text search point cloud retrieval method and device and electronic equipment, and the method comprises the steps: projecting a three-dimensional point cloud into an aerial view, and generating a BEV image; encoding the image by using a pre-trained vision-language model to obtain BEV feature vectors, storing the BEV feature vectors in a vector database, and constructing a point cloud feature library; meanwhile, encoding a user query text to obtain a text feature vector; searching a matching scene from the feature library by calculating the cross-modal similarity between the text and the BEV feature vector; and finally, mapping a retrieval result to original point cloud data and outputting the original point cloud data. By means of the method, the feature alignment problem of the point cloud and the natural language is solved, and the retrieval method which directly and efficiently utilizes essential features of the point cloud data to be in seamless joint with the natural language is achieved.

This application provides a data processing method, apparatus, device, and medium. The method includes: acquiring a first text containing business text data; performing a risk assessment on the first text to obtain a risk category result corresponding to the first text; if the risk category result is a first risk category, acquiring keywords of the business text data and searching for the keywords in a standard database; if a request text matching the keywords is found in the standard database, determining the feedback text corresponding to the request text as the business processing result corresponding to the business text data; if no request text matching the keywords is found in the standard database, performing text search processing on the business text data in a target knowledge graph to obtain a business processing result matching the business text data. Implementing this application embodiment can improve the security of text data.

The invention relates to a multi-line text searching and replacing method and system, and the method comprises the steps: carrying out the preprocessing of an original multi-line text, a multi-line searching content and a multi-line replacing content, removing the head and tail spaces of each line, and obtaining a corresponding logic text, a logic searching content and a logic replacing content, judging whether the logic search content contains a preset wildcard character or not, if the logic search content does not contain the preset wildcard character, carrying out simple character string matching on the logic text to obtain a logic index search result, converting the logic index search result into an original character index by utilizing a preset index mapping relation for positioning, and executing simple replacement operation, and if the logic search content contains the preset wildcard character, carrying out simple character string matching on the logic text to obtain a logic index search result; and if the logic text is matched with the fuzzy item list, regarding the logic search content as an integral matching block, performing sliding comparison according to rows in a row array corresponding to the logic text, if the matching succeeds, extracting the fuzzy item list and the starting and stopping row indexes of the matching block, performing positioning by utilizing an index mapping relation, and executing fuzzy replacement operation of inserting the fuzzy item list into a placeholder.

This invention provides a method and apparatus for generating tokens using computing power acceleration on an intelligent computing cloud platform. It relates to the fields of intelligent computing centers, smart computing centers, computing infrastructure, and smart computing cloud technologies. The method for generating tokens using computing power acceleration on an intelligent computing cloud platform includes: Step S1, establishing a full-text index of a target data set; Step S2, responding to a received data query request, determining whether to enable document index acceleration; Step S3, if document index acceleration is enabled, extracting scalar filtering conditions and vector filtering conditions; Step S4, performing a full-text search based on the scalar filtering conditions to obtain a first data set; Step S5, performing a vector search on the vector index of the first data set based on the vector filtering conditions to obtain a matching second data set, and returning it to the user. This invention can improve the speed of vector database processing with scalar constraints on intelligent computing cloud platforms, reduce resource consumption, and increase token generation speed.

Embodiments of the present application provide a text search method and device, electronic equipment and storage medium, and relate to the technical field of retrieval. First, according to a plurality of word data, an index library and preset search information, all texts containing the preset search information are filtered to obtain at least one first text; then, the at least one first text is sorted to obtain at least one second text; and according to the category and content of the second text, label information corresponding to each second text is generated; finally, the identification and label information corresponding to each second text are taken as search results and displayed. According to the category and content of the filtered texts, the corresponding label information is generated, and the texts and the label information are taken as the search results for display, thereby saving storage space, and text search can be realized without large equipment.

The invention discloses an article generation method and device, electronic equipment and a storage medium, and belongs to the technical field of content generation. The method comprises the steps of obtaining multiple document contents based on a specified article theme and / or keyword; generating an initial article based on the article theme, the keyword and the plurality of document contents by utilizing a first large language model; determining positions where illustrations need to be added in the initial article and illustration description information corresponding to the positions by utilizing a second large language model; searching the illustration in an image-text search material library based on the illustration description information to obtain a first search result, and determining a target illustration based on the first search result; adding a target illustration at the position of the initial article; and generating a target article based on the initial article added with the illustration. Therefore, technologies such as a large language model and image-text search can be combined, and the SEO article with high quality and rich illustrations can be quickly produced.

This invention discloses a method for implementing file tags based on a distributed file system, comprising: building a distributed file system on a multi-node server; establishing a file dimension tag information group storage facility, building a database to store tag group data, and periodically synchronizing the data to a full-text search engine to improve the retrieval speed of tag group information; constructing tag library rules, creating corresponding custom tag features on the distributed file system management platform by writing relevant programs for custom tag dimension management, and realizing the definition and management of tag dimensions for distributed file system files; and building an automatic tag identification capability for distributed system files, automatically identifying tags for distributed file system files based on pre-set word segmentation rules by writing relevant code for file tag setting logic, thereby meeting users' needs for classifying, managing, retrieving, collecting, and sharing distributed file system files.

The invention discloses an electric power safety regulation cross-text retrieval method and system based on graph retrieval. According to the method, firstly, an electric power regulation document is preprocessed, a knowledge data set is constructed, an entity relation graph and an intra-sentence co-occurrence hypergraph are synchronously constructed according to user problems, and a unified fusion graph matrix is formed through fusion; then, under the constraint of a preset token budget, an information-path collaborative sub-graph retrieval algorithm is adopted, and logically coherent evidence sub-graphs are accurately screened out from the fusion graph; the sub-graph is coded through a graph neural network, multi-dimensional features such as nodes, texts, topologies and types are fused, and a graph-level semantic vector is generated to serve as a structured soft prompt. And finally, splicing the soft prompt and text information, inputting the spliced soft prompt and text information into a parameter-frozen large language model, and driving the model to generate an accurate answer and a clear reasoning path. According to the method, the problems of evidence fragmentation and inference chain incompleteness in multi-hop questions and answers in the field of electric power security are effectively solved, and the accuracy, reliability and interpretability of answers are remarkably improved.

The application provides a knowledge system construction method based on text minimum information units. The text is subjected to keyword sorting based on full-stack full-text search technology, and the sorted keywords are subjected to auditing through text source auditing rules to determine a text term library; the relevance between terms is judged according to the text term library, and a term node network is established based on the relevance; a multi-text network is established according to the term node network, and a network knowledge system is generated. The application has the beneficial effect that the application understands the definition of key concepts in the text through keyword interconnection, so as to summarize the core content of a book as quickly as possible, and realizes super-fast reading; the term cloud of the application can stimulate creativity, produce relevance between two adjacent but seemingly irrelevant terms, and inspire new knowledge of innovation.

The present application belongs to the technical field of natural language search processing, and particularly relates to a short text search similar long text method without labeled data. That is, according to a batch of original long texts without labeling, sentence division, keyword extraction, long-short sentence mapping relationship establishment, and long-short sentence relationship pair as input source are performed; long text encoding and short text encoding are respectively performed on the long-short sentence relationship pair, and text feature representation is respectively obtained; model learning is performed, and a CLIP means is used to make the cosine similarity of a batch of training data existing long-short sentence mapping relationship maximum, otherwise as small as possible, and the cosine similarity of the data not existing long-short sentence mapping relationship minimum; after saving the model each time, the data is shuffled, and the probability of occurrence of negative samples is increased. The main innovation points of the present application are as follows: the keyword extraction and the CLIP architecture-based contrast learning technology are combined, a search mode based on deep learning semantic representation without labeling of user data is realized, and gMLP is used as a text encoder, so that the gMLP can be effectively used for knowledge retrieval.

1. Name of the product in this design: Graphical User Interface for Input Methods of Electronic Devices. 2. Intended use of this design: for use in electronic devices. 3. The key design features of this product are the graphical user interface content. 4. The picture or photo that best illustrates the key design points: Design 1 front view. 5. Design 1 is designated as the basic design. 6. Purpose of the graphical user interface: to display input method information. 7. Human-computer interaction method of graphical user interface: When entering text search, the main view of Design 1 is displayed. After clicking the input method button in the main view of Design 1, the interface change state diagram of Design 1 is displayed. When entering the text search interface, dragging the input method border can move it on the interface, and the main view of Design 2 is displayed.

The application provides a kind of multi-dimensional permission dynamic control method and system based on full-text search, method includes: the permission requirement mark of each business data object in configuration information system is configured;Each permission requirement mark is converted into permission feature string, as full-text search field is associated with business data object;In the process of information system operation, the permission matching log in user access request is collected, and the permission heat of each permission feature string is counted;According to the permission heat, each target permission feature string is identified as a permission hotspot set;The index item corresponding to the data object belonging to the same permission hotspot set is merged into the same index shard;According to the feature information of target access user, generate query condition, execute full-text search on the updated index shard structure, get the access data allowed by target user, can accurately identify and adjust permission hotspot, ensure the flexibility and efficiency of data access.

The invention discloses a text retrieval method and electronic equipment. The method comprises the following steps: acquiring an initial retrieval text; calculating the information entropy of the initial retrieval text according to the occurrence frequency of each vocabulary in the initial retrieval text; according to the information entropy of the initial retrieval text and a preset entropy interval, performing normalization processing on the information entropy to obtain a weight coefficient; determining the weight coefficient as a retrieval weight corresponding to the vector retrieval mode, and determining a retrieval weight corresponding to the full-text retrieval mode according to the retrieval weight corresponding to the vector retrieval mode; respectively retrieving the initial retrieval text in a vector retrieval mode and a full-text retrieval mode to obtain a plurality of retrieval result texts and retrieval similarity corresponding to each retrieval result text; determining a retrieval score of each retrieval result text according to the retrieval similarity and the retrieval weight corresponding to each retrieval result text; and according to the retrieval score of each retrieval result text, determining a target retrieval result corresponding to the retrieval request from the multiple retrieval result texts.

The invention provides a text retrieval method and a text retrieval system. The method comprises the following steps: dividing a software defect document to obtain a plurality of text blocks; generating a complementary vector corresponding to each text block according to the plurality of text blocks, wherein the complementary vector comprises a text block vector, an abstract vector and a context vector; generating a text block vector of the retrieval text according to the retrieval text; calculating a weighted average value of cosine similarity of the text block vector of the retrieval text and the complementary vector of each text block to obtain a similarity score of each text block; and forming a retrieval result by the text blocks of which the similarity scores are greater than the preset score. The problem that the retrieval effect is poor due to the fact that a single text vector is difficult to comprehensively express semantics in the prior art is solved.

A method and related device for text search processing are disclosed. This method searches for keywords in multiple runtime path information using keyword rules from a keyword rule set, resulting in relatively accurate search results that meet user search needs. The method includes: obtaining first text, which includes one or more runtime path information; obtaining a preset search rule set, each preset search rule indicating a logical relationship between at least one keyword; searching for one or more second keywords based on the first keyword and the first preset search rule, where the first keyword is a keyword obtained from the first runtime path information, and the first runtime path information is any one of the one or more runtime path information; and determining a first search result based on the first keyword and the one or more second keywords. The method provided in this application can be applied to log text processing in electronic devices such as smart cars, terminals, and computers.

The embodiment of the present specification provides a full-text search method and device in a block chain, which is applied to a service program deployed on a block chain service platform; including: obtaining a business model configured for each block of the block chain; wherein each block of the block chain stores business data with business semantics; the business model is used for data analysis of transaction data stored in each block of the block chain according to the business semantics to obtain the business data; based on the obtained business model, the transaction data stored in each block of the block chain is respectively analyzed, and a query index is established for the business data obtained by data analysis; the business data with the query index is sent to a full-text search engine connected with the block chain service platform, so that the full-text search engine executes data query on the business data based on the query index.

The invention belongs to the technical field of digital scheduling communication, and relates to a scheduling communication system audio and video retrieval method and system based on voice content recognition and a readable medium, and the method comprises the steps: converting an unstructured audio and video file into text data, marking the timestamp of each word or each sentence in the text data, and generating the text data with the timestamps; performing text optimization on the text data with the timestamp to generate a retrieval tag; original audio and video recording files, text data, timestamps and retrieval tags are integrated and stored into a full-text search engine, and an inverted index is established; and carrying out audio and video retrieval in the full-text search engine by inputting keywords. According to the method, the user can directly and quickly position the recording or video fragment containing the keyword by inputting the keyword; the fundamental spanning from metadata-based retrieval to voice content-based retrieval is realized, and the retrieval efficiency is greatly improved.

This invention discloses a real-time anomaly detection system for large-scale data environments, relating to the field of network data monitoring technology. The system implements a streaming process in feature extraction and classification prediction models. When generating feature traffic data, CICFlowMeter introduces a message middleware and proposes a master-slave architecture using Elasticsearch and MySQL for massive data anomaly retrieval. It combines the high-performance full-text search of Elasticsearch with the highly reliable persistent storage of MySQL, leveraging the advantages of each search storage engine. A dedicated synchronization module is designed for data update synchronization, and a Kubernetes server is built. After updating the classification model, Kubernetes performs rolling updates to ensure the business is always online. To prevent excessive traffic congestion and avoid traffic loss in the message queue, Kubernetes can also monitor excessive system CPU usage (i.e., excessive anomaly detection traffic data) and excessive anomaly detection thread usage, enabling horizontal migration of the system to ensure no data loss and improve system real-time performance.

The invention discloses a full-text retrieval method and device, a medium, equipment and a computer program product, and the method comprises the steps that a data set is obtained, a word segmentation model is constructed according to the data set, the word segmentation model carries out word segmentation processing on data according to a word segmentation length range, and the word segmentation length range is determined according to the minimum word segmentation length and the maximum word segmentation length; the minimum word segmentation length and the maximum word segmentation length are dynamically adjusted according to a first parameter and a second parameter in the word segmentation model; performing word segmentation processing on each piece of data in the data set based on the word segmentation length range through the word segmentation model to obtain a word segmentation segment of each piece of data under each word segmentation length, and constructing a data index under the corresponding word segmentation length according to the word segmentation segment of each piece of data under each word segmentation length, the data index is used for data full-text retrieval. Therefore, the data index of the data under various word segmentation lengths can be constructed, and the retrieval efficiency and the retrieval accuracy of the disordered data are effectively improved.

The application discloses a multi-round question and answer large model multi-path retrieval question and answer parameter and data source optimization method, which adopts a small model to recognize the intention of user query, so that better recognition performance is realized under the condition that the computing resources are limited. According to the recognized intention category, a large model is used to determine which data source in a vector database, a full-text search engine or a structured database is used for enhanced retrieval, while each query parameter is given in different application scenarios. Moreover, the intention recognition result is dynamically adjusted with different queries and historical question and answer of each round of user in multi-round question and answer, so that the final large model calling parameter and data source are formed. Compared with the prior art, the application realizes efficient intention recognition to obtain intention classification by using a small model under the condition that the computing power is limited, can select a suitable knowledge base and question and answer parameter configuration for a large model in various application scenarios, and has obvious advantages in low resource consumption and question and answer accuracy.

The invention relates to a text and image searching method based on an improved CLIP model, and belongs to the technical field of image processing. According to the invention, an image and related data thereof are collected and preprocessed; an improved CLIP network is constructed; training the improved CLIP network by using the preprocessed image and related data thereof; and performing text picture search by using the trained improved CLIP network. The method is suitable for video structuring and other tasks suitable for text image search, the retrieval accuracy index of text image search can be improved by improving the CLIP model, and the training speed of the model can be improved. The method can be applied to wide video monitoring scenes such as smart cities and smart parks, the corresponding picture result is quickly retrieved in the database based on the text content of the user-defined attribute input by the user, and the retrieval efficiency and usability are greatly improved.

The search system (10) includes: a search term acquisition unit (111) configured to acquire a search term; a full-text search unit (112) configured to perform a search operation based on the search term; and a visualization unit (114) configured to display a correspondence between the search term and a result obtained by performing the search operation.

The invention relates to a method for improving full-text search accuracy, and belongs to the field of text processing. The method comprises the following steps: performing data preprocessing on a to-be-retrieved document; recording a search log, and analyzing the search log; and automatically maintaining the service dictionary and the new word discovery limit list according to the update of the service content, and automatically initiating an asynchronous re-indexing task after the service dictionary is adjusted and published to re-establish an index for the existing file. According to the method disclosed by the invention, the accuracy and the search efficiency of full-text search can be enhanced by pre-displaying new words, analyzing search logs and the like.

The invention discloses a full-text retrieval method and device, equipment, a storage medium and a program product, and relates to the technical field of full-text retrieval. The method comprises the following steps: displaying a meta-model acquisition interface in response to a first input of an acquisition meta-model; in response to a second input of selecting a target data source, a target data table and a target data filtering script on the meta-model acquisition interface, acquiring and displaying field information of the target data table of the target data source; in response to configuration input of full-text retrieval attributes in the field information, determining a target field; in response to a third input used for obtaining field data of a target field, querying field data corresponding to the target field and meeting a target data filtering condition in a target data table of the target data source based on the target query statement; determining full-text retrieval data for full-text retrieval based on the target fields of the target data table and the field data corresponding to the target fields; and in response to the full-text retrieval request, performing full-text retrieval in the full-text retrieval data.

A kind of multi-modal tourism information positioning type retrieval method based on tourism knowledge graph, according to the multi-modal data in the mixed database of picture and travelogue and tourism video, construct tourism knowledge graph with weight, and save the semantic position index of data source of entity and inter-entity relationship in the process of construction and update, search entity and inter-entity relationship are extracted from text when user carries out text search, mapping to a subgraph of knowledge graph, according to the corresponding index, return the retrieval result after the subgraph is enhanced and expanded.This application returns the result of retrieval text, which is also multi-modal, and points to the corresponding position of semantics.For travelogue data in database, return the text and picture corresponding to enhanced subgraph and the travelogue;For tourism video data in database, return the video clip and entire video corresponding to enhanced subgraph.The application solves the problem that multi-modal data is difficult to manage effectively, and tourism data retrieval is difficult to locate to target semantic unit.