7 results about "English characters" patented technology

This invention provides a method for extracting and restoring Chinese and English text and Arabic numerals from Uyghur PDF documents. It utilizes PDFMiner to extract all element information from the PDF document, thereby extracting all text content. Then, all text is sorted line by line and formatted from right to left. All Chinese and English characters, Arabic numerals, and other text characters sorted from left to right are selected and then sorted again from left to right. Finally, all characters are inserted into a Word document according to their corresponding format, thus ensuring a higher degree of consistency between the reconstructed Uyghur document and the original document. This invention can automatically detect and extract characters represented from left to right in Uyghur PDF documents and then insert them into the corresponding positions in a DOCX file in a left-to-right order, thereby ensuring that the layout of the target document is as consistent as possible with the original document.

The application provides a mixed language text processing method and device, electronic equipment, medium and product, and relates to the technical field of natural language processing. The mixed language text processing method comprises the following steps: obtaining a mixed language text to be processed; separating a target English string from the mixed language text to be processed based on a preset language distinguishing rule; and performing recursive subword segmentation processing on the target English string based on a subword vocabulary to obtain a target English subword sequence. The application can improve the accuracy of mixed English text segmentation and the accuracy of English vocabulary generation.

An English spelling correction method encoding each English character of each input English word string and normalizing a corresponding encoding number to obtain an initial numbering array numeric, performing distance calculation and normalization on the encoding number to obtain an initial distance array numeric, using a number (n) of English characters in an input English word string to generate a feature map with a variable length, performing a one-dimensional convolutional neural network operation and a fully-connected layer on the feature map to generate a feature map, outputting the feature map by a bidirectional long short-term memory (LSTM) network and performing a fully-connected layer to generate a feature map, performing a probability recognition and outputting a probabilistic word string, recognizing the probabilistic word string and outputting an English spelling correction result.

The present application belongs to the field of Braille translation, and particularly relates to a Braille translation method and device based on rules and multi-task fine-tuning and a storage medium. The method comprises the following steps: obtaining Chinese corpus, constructing a Braille rule engine, and cleaning, segmenting, pinyin annotating and Braille transcribing the Chinese corpus to generate parallel corpus containing Chinese, pinyin, Braille sequence and component structure; extracting Braille word units and candidate meanings from the parallel corpus and rule files, constructing a rule priori library, adding Braille word units to a model segmenter, and initializing new word unit embedding with anchor points such as pinyin, Chinese characters, numbers, punctuation and English characters; constructing instruction samples of Braille to pinyin, component analysis, pinyin to Chinese and Braille to Chinese, and performing course multi-task fine-tuning on the priori enhanced model to obtain an optimized translation model. The present application can improve the accuracy of Braille translation, provide Braille component analysis function, reduce the cost of manual transcription and correction, and is suitable for reading assistance, teaching correction and text digitization.

The invention belongs to the technical field of VR text input, and discloses a self-adaptive VR text input method based on rocker operation, which comprises the following steps of: selecting two English characters from 26 English characters of a full keyboard as anchor characters; dividing the 26 English characters into two character sets based on the position relation between the full keyboard English characters and the anchor point characters; by taking the two anchor point characters as the circle centers of the rocker direction operation areas of the two VR controllers, mapping and arranging the rest English characters in the two character sets in different directions of the circumferences of the corresponding rocker direction operation areas so as to form a double-ring-shaped virtual keyboard; according to the double-ring-shaped virtual keyboard, rocker input logic is constructed, and the rocker input logic is used for mapping operation executed by a user on a rocker into a corresponding input result; the problems of low input efficiency, high fatigue degree and poor input accuracy caused by the fact that a user needs to accurately search and click keys on a virtual full keyboard during VR text input in the prior art are effectively solved.

This application provides a watermarking method, an electronic device, and a computer-readable storage medium, relating to the field of digital watermarking technology. The method includes: acquiring a file to be processed and a first watermark key; parsing the file to be processed to extract target text lines containing English characters; adjusting the text layout parameters of the target text lines according to the first watermark key to obtain adjusted layout parameters, the adjusted layout parameters including the target width of each English character in the target text line; and generating a target file with watermark information based on the adjusted layout parameters of the target text line, the watermark information being represented by the target width of each English character in the target text line. This application achieves invisible watermark embedding by modifying the width of English characters in the file to be processed, subtly encoding the watermark information into the character width, thus not affecting the readability and format compatibility of the file, while achieving an efficient and covert copyright protection mechanism.

The application relates to a business semantic recognition method, which comprises the following steps: step S1, receiving an English character string and other context information of a to-be-recognized semantic; step S2, pre-processing the to-be-recognized field; step S3, segmenting the to-be-recognized field to obtain a plurality of segmentation schemes; step S4, for each segmentation scheme, using a transition probability matrix to calculate a probability score at a word segmentation point; step S5, for each word in the segmentation scheme, judging whether the word is a non-pinyin through a pinyin semantic DFA tree, and then performing one-by-one matching with a word entry in a word library and calculating a similarity score; step S6, taking a semantic of a word entry with the highest similarity score to calculate a correlation score between semantics of the whole matching scheme; step S7, extracting a matching scheme with the highest comprehensive score and splicing the matching scheme as a recommended label; and step S8, outputting a recommended Chinese business semantic and a recommendation index. The application improves the efficiency of metadata carding work.