74 results about "Table (information)" patented technology

The invention provides a table recognition method fusing multiple text features and geometrical information. The method comprises: a data processing step of obtaining a picture of a table area, and respectively carrying out OCR and straight line recognition on the obtained picture to obtain key feature information; a graph convolutional neural network training step of performing graph convolutional neural network training according to the obtained key feature information, and constructing a table structure recognition model; and a table identification step of performing structure identification on the table in the picture format according to the constructed table structure identification model. The invention provides thea table recognition method fusing the multiple text features and the geometrical information. Improvements are made in the aspects of diversity of adopted data, a method for performing feature extraction on the data and the like, the accuracy of table recognition is effectively improved, a more accurate table structure reconstruction result is obtained, and compared with an existing table recognition mechanism based on traditional rules and a traditional deep learning method based on pictures, the table recognition method has the advantage that the effect is greatly improved.

The invention relates to the artificial intelligence technology, and discloses a table structure identification method comprising the following steps: obtaining a to-be-identified image, and carryingout table area extraction and correction on the to-be-identified image based on semantic segmentation so as to obtain a table area graph; performing table frame line segmentation on the table area graph by using a pre-constructed segmentation model to obtain a table frame graph; performing image connected domain analysis on the table framework diagram to obtain table structure information; and performing text detection and identification on the table area graph to obtain a text detection result, and generating a spreadsheet document corresponding to the to-be-identified picture according to the text detection result and the table structure information. In addition, the invention also relates to a blockchain technology, and the to-be-identified image can be stored in a node of a blockchain.The invention further provides a table structure recognition device, electronic equipment and a computer readable storage medium. The accuracy of table structure identification can be improved.

The embodiment of the invention provides a government affair-oriented workflow management system and a library and table division method thereof, and the system comprise:s a data storage layer which is used for storing flow data and flow configuration data; a process engine which is used for providing any multiple services in process resource management, task scheduling and historical data query; an unified interface layer which is used for providing a plurality of unified interfaces; a service process extension module which is used for providing a government affair-oriented workflow management function; a process designer which is used for drawing a process model and configuring related information of process nodes; and a presentation layer which is used for managing the flow configuration information. The embodiment of the invention provides a government affair-oriented workflow management systemwhich provides the functions of design modeling, debugging, deployment, operation, monitoring and the like of a service process based on a workflow engine. High concurrent access of the engine is supported; by packaging and simplifying a workflow engine service interface, a unified API is provided for developers, and the process threshold of the developers is reduced.

The invention discloses a periodical literature table extraction method based on text state characteristics, and belongs to the technical field of PDF file information extraction. The method comprisesthe steps that a table model is constructed according to characteristics of tables in periodical literatures, the table model is composed of a table title and table content, a text line set is established, table detection, table data positioning and table reconstruction are carried out, a table cell data set is established, the table data is combined, a table cell data set is verified and the table data is output. The method is particularly designed for the characteristics of tables such as a three-line table widely existing in PDF periodical literatures. Accurate and correct extraction of specific tables in PDF periodical literatures can be achieved, particularly, the logic relation of three-line tables is guaranteed, manual intervention and interaction are not needed in the whole process, table selection is not needed, and automation of the whole extraction process can be achieved. As the ontology is adopted to help filter the table, the method also has wide applicability.

The invention discloses a table implementation method. The method comprises the steps of obtaining height and width information of a table content main body and height and width information of rows and columns; wherein the table is generated by a div label; determining the table row and column number of the current visible area of the table content main body; creating an anchor point row and endowing an initial height value, and creating an anchor point column and endowing an initial width value; monitoring a div rolling event for the anchor point row and the anchor point column to obtain a rolling distance value; updating row and column contents needing to be displayed in the visible region according to the rolling distance value; and rendering the row and column contents needing to be displayed in the visible region. According to the method, the boundary values of the visible region and the target region can be more easily calculated through the div label, the rendering capability ofthe dom node of the complex table can be efficiently realized by combining the characteristics of different box models, and the performance and the efficiency are higher; the div label can avoid theworkload of the fixed column function simulated by the table label in row and column width and height calculation, and the calculated amount is small.

The embodiment of the invention provides a table recognition method and device, a terminal and a storage medium. The method comprises the steps: acquiring a to-be-recognized file which comprises at least one table; utilizing a table position detection model to detect position information of at least one table in the file, the table position detection model being a deep learning model; processing the file according to the position information to obtain a table picture which corresponds to each table and only contains the table; carrying out the following processing on each table picture: performing pre-processing operation, processing the processed table picture according to a table row division model and a table column division model, and acquiring row position information and column position information of the table, wherein the table row division model and the table column division model are both deep learning models, so that the structure information of the table is restored; and outputting structured table data in combination with the recognized text information of the table. In the way, the table recognition efficiency and flexibility can be improved.

The invention relates to the technical field of financial science and technology (Fintech), and discloses a web page table setting method, which comprises the following steps of: when a table settingrequest is received, obtaining a table corresponding to the table setting request in a web page; generating a table setting window according to the information in the table and outputting the table setting window so that a user can input table configuration information based on the table setting window; and when it is detected that the table configuration information is input completely, adjustinga table in the web page according to the table configuration information. The invention further discloses a web page table setting device, equipment and a computer storage medium. According to the method and the device, the real-time performance and the convenience of web page table setting operation are improved.

The invention discloses a nested table extraction method and device and a storage medium, and the method comprises the steps: reading data contents in a first type of file, carrying out the analysis, and determining the line segment coordinate information contained in the analyzed data contents; grouping the line segments by taking the display unit as a reference; based on the coordinates of the line segments, merging the line segments of which the transverse or longitudinal distances of adjacent or connected line segments are smaller than corresponding thresholds and the adjacent endpoint intervals of the adjacent line segments are smaller than corresponding fixed thresholds; traversing all the merged line segments, determining whether the line segments are intersected or not, and generating a set; traversing the line segments in the set, determining intersection points between the merged intersection line segments, traversing all the intersection points, and determining intersection points forming a rectangular frame; generating a table by taking an intersection point of which the area of the formed rectangular frame is greater than a set threshold value as a reference; and calculating an inclusion relation of the table based on the generated table, and forming a nested table based on the inclusion relation. According to the invention, the form extraction accuracy is improved, and the editing work is reduced.

The invention discloses a table structuring method, a table recovery system and a device with a storage function. According to the table structuralization method, the deep learning scheme is fully utilized to obtain the position information and the text information of the table cells, and the table lines and the table cell fixed point position information are accurately mined, so that accurate table structuralization is obtained.

The invention provides a table reconstruction method and device, computer equipment and a readable storage medium. The method comprises the steps of obtaining a table image; detecting and identifying the table image to obtain text contents and layout information corresponding to a plurality of textboxes, and coordinates of row table lines and coordinates of column table lines; wherein the layout information comprises position information, a row sequence number and a column sequence number; and performing table reconstruction according to the text content, the layout information, the coordinates of the row table lines and the coordinates of the column table lines. According to the method, the row sequence number and the column sequence number corresponding to each text content are obtained, so that the table can be reconstructed by combining the row sequence number, the column sequence number and the detected table line, and compared with the prior art that the table is reconstructed by only depending on the detected table line. The method can accurately restore the actual structure of the table, and the accuracy of table reconstruction is improved.

The embodiment of the invention discloses a table style processing method and a related device, which are used for realizing that the file of the whole table occupies a smaller storage space under thecondition that the same content information is stored in the table. The method provided by the embodiment of the invention comprises the steps of obtaining a target table; determining a table basic style corresponding to the target table; receiving cell style setting of the target table to form a table target style; calculating a table differential quantity style of the table basic style and thetable target style to form a table differential quantity style file; and associatively storing the table difference style file and the target table.

The invention provides a method, a system and a medium for establishing a physical model for a power grid knowledge graph. The method comprises the following steps: determining a table mode for defining a table object and a field thereof; generating table information of all table objects according to a table mode based on the first data source to generate a physical table set of a physical model;determining a relationship mode for defining a relationship between the source table object and the target table object; for each pair of source table objects and target table objects subjected to deduplication processing in the second data source, generating corresponding table relationship information according to a relationship mode based on the second data source so as to generate a relationship set of the physical model; building a physical model including table objects, fields, and relationships based on the physical table set and the relationship set. By utilizing the scheme of the invention, knowledge extraction can be carried out on different data sources, and leakage detection and vacancy filling are carried out on the existing model to make up the design weakness of the existingmodel, so that a more reasonable management and control model is provided for a user, and information matching of a unified data model is supported.

The invention relates to a table extraction method and device, computer equipment, a storage medium and a computer program product. The method comprises the steps that a table image is obtained, the table image is input into a table extraction model, the table extraction model comprises a text extraction network block, a cell extraction network block and a structure extraction block, text extraction is conducted on the table image through the text extraction network block, and text information in cells in a table area of the table image is obtained; performing cell extraction on the table image through the cell extraction network block to obtain cell information of cells in the table area of the table image, generating a table structure corresponding to the table area through the structure extraction block according to the cell information of the cells, and fusing the text information with the cells according to the table structure to obtain the spreadsheet. By adopting the method, end-to-end table data extraction can be performed on the table image by utilizing the table extraction model, the structured spreadsheet is generated, and the universality of the table extraction method is improved.

The invention discloses a table frame prediction model generation method and device and a table positioning method and device, and the method comprises the steps: obtaining a plurality of sample document pages containing tables, and converting the sample document pages into sample images; dividing each sample document page and each sample image into S * S first grids and S * S second grids respectively; obtaining semantic information of texts contained in the S * S first grids as semantic information of the S * S first grids; obtaining prediction frame information of S * S second grids; and taking the obtained information as a training sample set, iteratively training an image recognition model, and generating a table frame prediction model for predicting table frame information. The imagerecognition model adopts a large number of training samples and fuses semantic information in the training process, so that the table frame information can be better predicted, the success rate of table frame determination is improved, and the smooth proceeding of the subsequent table positioning process is ensured.

The embodiment of the invention provides a table-based retrieval method and device . The method comprises the steps of conducting retrieval according to a slot position corresponding to an initial problem to obtain a retrieval result; judging whether the number of the retrieval results exceeds a threshold value or not; if the number of the retrieval results exceeds a set threshold value, performing at least one interactive retrieval step; wherein the interactive retrieval step comprises the steps of determining unused slots according to slot sorting information, generating an inquiry statementby utilizing the unused slots, collecting interactive problems according to the inquiry statement, and combining the slots corresponding to the interactive problems with the used slots for retrievalagain to obtain a retrieval result. According to the embodiment of the invention, whether interactive retrieval is carried out or not can be determined according to the retrieval result, a slot collection process does not need to be fixed in advance, and the unused slot is determined according to the slot sorting information, so that multiple rounds of dialogues during table-based retrieval are more intelligent and flexible, and the problem solving rate of a user can be increased.

The invention discloses a table recognition method, a model training method and device and equipment. The table recognition method comprises the following steps: when cross-page broken line recognition is performed on a table, obtaining first text information from a first cell in a first page; determining a second cell corresponding to the first cell in the second page; then obtaining second text information from the second cell; performing natural language analysis on the first text information and the second text information; and when the analysis result is that the first page and the second page belong to the context statements, determining that the table has cross-page broken lines between the first page and the second page. According to the table recognition method, whether the table is truncated between the two pages or not is recognized by judging whether the text information of the table in the cells of the two pages belongs to the context statements or not, and the recognition precision of cross-page broken lines of the table can be effectively improved based on the logical coherence of the original text information of the truncated cells. The invention can be widely applied to the technical field of artificial intelligence.

The invention discloses a table picture analysis method and system, and the method comprises the steps: recognizing the text content and text position in a table picture, obtaining a triple set, whereeach triple consists of a text, a position horizontal coordinate and a position vertical coordinate; inputting the table picture into a segmentation model containing the trained convolutional neuralnetwork to obtain segmentation line information of the table; and combining the segmentation line information with the triple set according to a preset rule to form a structured table document. According to the method, the table picture can be automatically converted into the structured data, so that the labor cost and time are greatly saved.

The invention discloses a table information extraction model training method, which comprises the following steps: processing table cells of table corpus to obtain feature vectors of the table cells; calculating an adjacent matrix according to the position information of the cells, and performing feature extraction on the feature vectors of the cells and the adjacent matrix to obtain high-order feature vectors of the cells; predicting an original text of the cell by using the high-order feature vector, and performing model training by using the text of the cell to obtain a table language model; and training the table language model by using the training sample corresponding to the current table information extraction task to obtain a table information extraction model. For each different table extraction task, only a small number of training samples corresponding to different tasks need to be used for training to obtain the corresponding table information extraction model on the basis of the trained table language model, and the table language model does not need to be trained in each training, so that the training time and the training samples are obviously reduced.

The invention provides an analysis method for table line and page changing, which comprises the following steps: judging the definite line and page changing situation through the expert experience summary rule; The tagged corpus is obtained by using depth learning model. According to the tagged corpus and the training depth learning language model, it is judged whether two adjacent cells can be merged or not. The device for implementing the method comprises: a wrapping and page-changing condition judging module for judging a definite wrapping and page-changing condition through an expert experience summary rule; and a wrapping and page-changing condition judging module for judging a definite wrapping and page-changing condition. The tagged corpus acquisition module is used for acquiring the tagged corpus acquisition module of the tagged corpus by using the depth learning model. A cell merge judging module is used for judging whether two adjacent cells can be merged according to the tagged corpus and the training depth learning language model. The invention utilizes a depth learning model to mine semantic information contained in a table, and can accurately analyze whether two adjacent cells can be merged in a line and page changing scene.

The invention provides a human-computer interaction method, device and equipment based on table data. According to the method, the retrieval library is constructed offline, and the SQL statements of table data support query and the problem representation vectors corresponding to the SQL statements are stored in the retrieval library. During online question answering / dialogue, an input question is encoded to generate a semantic representation vector of the input question, and similarity matching is performed on the semantic representation vector of the input question and a question representation vector in a retrieval library, so that an SQL statement accurately matched with the input question can be found, and Text-to-SQL language conversion is realized; model training does not need to be carried out based on a data set in a specific application field, generalization is higher, and efficiency is higher; furthermore, the SQL statement corresponding to the input question is operated, the table data can be inquired to obtain the answer information of the input question, the response time during online question answering / dialogue can be shortened, and the response timeliness of the human-computer interaction system based on the table data is improved.

The invention relates to a list table generation method and device, computer equipment and a storage medium. The method comprises the steps: when a list table generation instruction is received, obtaining a corresponding target list based on control information contained in the list table generation instruction; Obtaining target data information contained in the target list, and identifying and determining a category attribute corresponding to the target data information; And generating a list table corresponding to the target list based on the category attribute corresponding to the target data information. By the adoption of the method, the target list can be completely exported to generate the corresponding list table, the data information in the obtained list table is stored with the corresponding incidence relation, and format setting does not need to be conducted on the list table again.

The invention discloses a lightweight text labeling method and system based on javascript, equipment and a storage medium. The method comprises the following steps of analyzing an SQL statement; obtaining a table name array according to an analysis result of the SQL statement; adding a span label to each table name in the table name array; and obtaining label position information of the span label, and obtaining the position of the table name in the SQL statement according to the label position information. Through the method and the device, the technical problem that the position of a table cannot be conveniently found after the super-long SQL statement is analyzed can be solved, and the method and the device are simple to use and high in expansibility.

The invention relates to a table recognition method and device, a readable medium and electronic equipment, and the method comprises the steps: obtaining a table image, determining table structure information according to the table image through a table recognition model which is used for obtaining a text semantic graph corresponding to the table image, and determining the table structure information according to the table image and the text semantic graph; and determining a multi-modal feature graph, detecting each cell in the table image according to the multi-modal feature graph to obtain a detection position of each cell, determining a node feature corresponding to each cell according to the detection position and the multi-modal feature graph, and determining table structure information according to the node feature. According to the table structure identification method and device, the table structure identification is carried out through the table identification model by utilizing the multi-modal node features, accurate table structure information can be obtained, the identification effect on different types of tables is ensured, and meanwhile, the universality of table structure identification and the accuracy of the generated target table are improved.

The invention provides table entry management method, device and equipment and a machine readable storage medium, and the method comprises the steps: obtaining a to-be-inserted table entry group whichcomprises to-be-inserted table entries needing to be inserted into a pre-established access control list; obtaining an insertion position of each to-be-inserted table entry of the to-be-inserted table entry group; adjusting the access control list according to the insertion position of each to-be-inserted table entry of the to-be-inserted table entry group; and inserting the to-be-inserted tableentries of the to-be-inserted table entry group into the access control list. According to the technical scheme, the information of the multiple to-be-inserted table entries of the to-be-inserted table entries is obtained, the existing table entries in the access control list needing to be adjusted are migrated at a time, then the multiple to-be-inserted table entries are inserted at a time, and therefore the migration amount of the table entries in the access control list is reduced, and the problem that the migration amount of the table entries is huge is solved.

The invention discloses a data migration method and system, and the method comprises the steps: obtaining data information of a source system; obtaining a target system; analyzing the target system, and dividing the data information into first static data and first dynamic data; mapping the first static data to the target system to obtain second static data; splitting the processing logic of the target system to obtain a plurality of sub-processing logics; according to the first dynamic data, enabling the static data mapped to the target system to be subjected to batch processing through the multiple sub-processing logics, obtaining second dynamic data, wherein the second dynamic data are the same data as the first dynamic data in the target system. The technical problem that in the prior art, the difference between a new system and an old system is large, if a traditional library table mapping mode is adopted for data migration, large uncertainty exists, and consequently data cannot be directly mapped to a new system is solved.

The invention provides a graph for displaying the completion condition of a pipe jacking project based on table editing software, which comprises a title part, a completion condition part of a pipe jacking working well, a completion condition part of a pipe jacking, a remark part and an overall information part which are sequentially distributed from top to bottom, wherein the graph for displayingthe completion condition of the pipe jacking project based on the table editing software is manufactured through the table editing software, and the table editing software comprises Excel; on one hand, the completion condition of each part of the pipe jacking engineering can be visually displayed on the drawing, and the requirement of engineering managers for mastering actual condition decision reference is met; and on the other hand, the manufacturing of the pipe jacking technology schematic diagram is easier and more convenient, the manufacturing can be easily completed by adopting table editing software, and convenience is provided for non-professional drawing personnel to manufacture the schematic diagram of engineering management.

The invention belongs to the technical field of artificial intelligence and provides a text duplicate checking processing method, a device, computer equipment and a computer storage medium. The methodcomprises the steps of obtaining the similarity between a text to be duplicated and a historical text by obtaining a word score table corresponding to the text to be duplicated, combining a word score value corresponding to a target word in the word score table and performing comparison calculation on the text to be duplicated and the historical text stored in a historical text database accordingto a word level; comparing the similarity with a similarity threshold, and evaluating whether the to-be-duplicated text is a duplicated text or not according to a comparison result. According to themethod, similarity between texts is compared and calculated based on word score values corresponding to words; according to the text duplicate checking method and device, words containing special information have large influences on text similarity evaluation, general words have small influences on text similarity evaluation, whether the content of the two texts is strongly related to the words inthe two texts or not is judged repeatedly, and the accuracy and reliability of text duplicate checking are improved.

The invention discloses a table description information generation method. The method comprises the steps that to-be-processed table name information in a metadata table is acquired; preprocessing the table name information to obtain a word sequence which belongs to a first language; based on the word sequence, text probability distribution is obtained through a text generation model, and the text probability distribution comprises at least one word probability distribution; table description information corresponding to the table name information is generated according to the text probability distribution, the table description information comprises at least one word, each word in the at least one word corresponds to one word probability distribution, the table description information belongs to a second language, and the second language and the first language belong to different languages. The invention further provides a device, equipment and a medium. According to the method, the text generation model is adopted to convert the table name information, and the table description information corresponding to the table name information can be automatically complemented without manual participation, so that the labor cost is reduced, the working efficiency is improved, and normal operation of businesses is facilitated.

The invention relates to a table extraction method and device in an engineering drawing, electronic equipment and a storage medium, and is applied to the technical field of data processing. The method comprises the steps of determining a to-be-recognized area in an engineering drawing, wherein the to-be-recognized area comprises first elements in a to-be-extracted table, and then position information of all the first elements in the to-be-recognized area is acquired; according to the position information, determining a position relationship among the first elements; and storing the first element in a table form according to the position relationship to obtain a to-be-extracted table in the engineering drawing. In the prior art, a line splicing mode is adopted to classify, splice and sort structured data of a table. However, due to the fact that the number of lines in the architectural drawing is large and the time complexity of splicing the lines is high, the algorithm is long in consumed time and low in efficiency.