98results about How to "Improve translation accuracy" patented technology

The present solution includes a method for dynamically translating application prompts to internationalize software applications for a non-native language that is not specifically supported by the application. In the solution, application prompts can be identified that are associated with a software application. Each application prompt can include text written in an original language. An attempt of the software application to render one of the application prompts can be intercepted and dynamically translated. The translated text can be substituted for the original text. The application prompt can then be rendered.

The invention provides a neural machine translation method of merging multilingual coded information, and relates to a method of neural machine translation. The purpose is to solve the problem of low accuracy of translation in the prior art. The process comprises the steps of firstly, obtaining subcharacter symbol sequence corresponding to each language, establishing dic_s1,dic_s2 and dic_t; secondly, inputting a word vector into an NMT model for training, updating the word vector according to an initial value training, until the bleu value of the NMT model is increased by 1-3 points; thirdly, obtaining ctx_s1 and ctx_s2; fourthly, obtaining the merged result; fifthly, obtaining C; sixthly, calculating qt+1 at time t +1 according to the formula to obtain the probability distribution pt+1 of the word y't+1 at time t +1 for a target language sequence, sampling the target word y't+1 at time t +1 according to pt+1, until the closing tag of the sentence is decoded, and the decoding translation is finished. The neural machine translation method of merging multilingual coded information is used in the machine translation field.

The invention belongs to the technical field of multimedia information processing, specifically relating to a Chinese-English unknown words translating method blending Web excavation, multi-feature and supervised learning. The method provided by the invention comprises four steps as follows: pre-processing and estimating parameters; acquiring translation candidates based on Web; representing the translation candidates based on multi-feature; ordering and estimating translation results. In the invention, a method for acquiring corpora through Web excavation, representing the translation candidates with multi-feature, and ordering the translation candidates through supervised learning. Compared with the traditional method, the method provided by the invention has the advantages of simple corpora acquiring method and pre-processing, overall features for representing the translation candidates, and high accuracy for translation results. Unknown words translation is one of important and difficult points for text processing; the effective Chinese-English unknown words translating method is provided by the invention, so that the method has the important application value in machine translation and cross-language information retrieval fields.

The application provides a method for long statement segmentation aiming at neural machine translation. The method comprises the steps that before statement translation based on an NMT model, direct input of source statements into the NMT is replaced by segmentation of statements into short sub-statements, and each sub-statement is successively input the NMT model, so that each segmented sub-statement is translated successively by the NMT model respectively; and then the translated sub-statements are directly spliced into a complete sub-statement. The sub-statements which are input into the NMT model for translation are short and translation accuracy of the NMT model is high, so that the accuracy for statement translation is increased. In addition, the application also provides a device for the long statement segmentation aiming at the neural machine translation so as to ensure application and implementation of the method in practice.

The invention discloses a system and method for translation. The method comprises the steps that an original text needing to be translated is input, cut into pieces, numbered and displayed on a text frame; a language database is searched for the original text pieces, the original text pieces are matched with texts in the language database, the translation texts highest in matching degree are selected in a default mode, combination is conducted according to the numbers, and the combined translation texts are displayed on a translation text frame; the translation texts in the translation text frame are manually checked, the corresponding number of the target translation text needing to be modified is clicked, an expansion frame pops up, and replacement or modification is conducted according to a matched translation text corresponding to the number and displayed on the expansion frame. According to the system and method, checking and correction are conducted through the method that the matched translation text is displayed in a window, the accuracy of translation is improved, labor cost is saved, and the experience comfort degree of a user is improved.

The embodiment of the invention discloses a text translation method and device, equipment and a storage medium. The method comprises the steps of obtaining an original language text; translating the original language text by using a text translation model; obtaining a target language text corresponding to the original language text, wherein the text translation model is a translation model obtained by correcting an original translation model according to a text evaluation result of a training translation, the training translation model is an output result after translation of the original translation model, and the text evaluation result is used for evaluating the semantic relationship of the context text in the training translation. According to the technical scheme provided by the embodiment of the invention, the problem that in the prior art, semantic consistency before and after translation of a translated text obtained by independently translating each sentence is poor is solved;according to the technical scheme, the text translation model is effectively corrected, the translation accuracy of the text translation model is improved, and then the front-back semantic consistencyand fluency of translated texts are improved.

The invention relates to a multimedia-based, multi-platform, multi-terminal, and multi-language translation method with high translation accuracy. The invention includes the following steps: (1) information classification and recording; (2) client terminals, after system initialization, can be used to select the language environment suitable for themselves. The language, called Language A, can be clicked to enter the interface; (3) the client terminals select the target language B in the suitable language environment; (4) information searching; (5) the client terminals can continue to select the target language C according to the multimedia information. The server terminals, according to the multimedia information, search the corresponding pronunciation and paraphrase of language C in the multi-language meaning database, and send the result to the client terminal through analysis program, thus completing the multi-language translation from Language A to Language B, and then to Language C based on multimedia. The invention can, with multimedia information as the medium for the multilingual translation, directly reveal the meaning of sentences to achieve the translation accuracy of information interactivity among multiple languages.

The invention provides an online translation method, a device, a system and a server. The online translation method includes: receiving an online translation request transmitted from a user terminal; acquiring communication language between two communicating parties; and performing online real-time translation for communication information carried in the online translation request. The communication language is carried in the online translation request. Compared with the prior art, the online translation method, the device, the system and the server have the advantages that communication information can be carried in the online translation request to be transmitted, manual copying of the communication information is omitted, and the online real-time translation and timely communication are realized simultaneously. Furthermore, the communication information can be translated by online translator, the translation accuracy is improved as compared with a mode that the communication information is copied to translation software to be translated, and the communication quality is guaranteed.

The invention relates to a method for determining meanings of a polysemous word. The method comprises the following steps of acquiring an association text of the polysemous word w of a meaning to be determined, and finding out feature words of the word w from the association text; sequentially judging the found feature words in a decision tree for determining the meanings according to a generation sequence of the decision tree; determining the meaning needed to be adopted by the word w currently according to a determining result. According to the method for determining the meanings of the polysemous word, the translation accuracy of the polysemous word under different styles and contexts is improved, the threshold of polysemous word text translation is reduced, the translation efficiency is improved, and manpower and material resources are greatly saved.

The invention belongs to the technical field of processing or conversion of natural languages, and discloses a machine translation method and machine translation systems for automatically recognizinga chapter based on fields. The machine translation method comprises the steps of dividing sentences to be translated to the corresponding fields; calling the translation systems corresponding to the fields for each sentence to be translated in the chapter according to corresponding field information to translate the sentences; merging the translated sentences to obtain a final chapter translationresult. On the basis of a traditional machine translation mechanism, improvement is conducted, a classifying algorithm is adopted to divide the fields of the sentences to be translated, and then the sentences are put into the translation systems corresponding to the fields for translation. Compared with a traditional method of manually selecting the translation systems of the fields for translation, the time for manually judging the fields to which the sentences belong can be saved. Compared with the mode of directly utilizing a universal translation system for translation, the translation accuracy can be effectively improved.

The present invention relates to the field of natural language processing and discloses a method and device for translating text information, a terminal device and a computer-readable storage medium.The method for translating the text information comprises: analyzing the text information to be translated to determine a subject text and format information of the text information to be translated;determining a plurality of subject frames of the text information to be translated or at least one subject element in each subject frame on the basis of the subject text and the format information; and performing sub-topic frame translation or sub-topic element translation on the subject text by using a trained translation model corresponding to each subject frame or each subject element. The method of the embodiment of the present application realizes the more fine-grained content extraction of the text information to be translated, can realize the targeted translation of the sub-topic framesor sub-topic elements of the text to be translated, so that the translation result has a clear subject and strong topic description characteristics, and the translation accuracy is improved.

The invention discloses a translation method, device and equipment based on machine learning, and a storage medium, and relates to the field of artificial intelligence, and the method comprises the steps: obtaining a sentence of a first language type; dividing the sentence into at least two word segmentation sequences by adopting different word segmentation devices; generating a word graph structure of the sentence according to the at least two word segmentation sequences; calling an encoder to convert the word graph structure into an intermediate vector representation of the sentence; and calling a decoder to convert the intermediate vector representation into a sentence of a second language type. Because the word graph represents the possibility of covering various word segmentation of the sentence, the problem that wrong segmentation is generated in word segmentation can be solved, the problem that wrong semantics or ambiguity is generated in the sentence due to the fact that wrongsegmentation is generated in word segmentation, and irreparable damage is caused to the semantics of the sentence is solved, and the translation accuracy of a machine translation model is improved.

The invention discloses a method for automatic identification of foreign language statements commonly used in a travel and corresponding scenes. The method comprises the steps that firstly, a commonly-used statement initiated by a user is received according to the user's translation request for the commonly-used statement, the statement translated or played by the user is reported to a server by a log, and the user's geographic information including a longitude and a latitude is acquired according to a GPS module during the log reporting; secondly, according to the user's geographic information including the longitude and the latitude, the information is sent to a POI data server, and POI information at the location is requested; thirdly, the POT data server returns the matched POI information, and a destination and a scene corresponding to the user's POI information are obtained; and fourthly, according to the destination and the scene of the user, a translation result corresponding to the translation request initiated by the user for the commonly-used statement is matched for the user. The method disclosed by the invention can be used for the automatic and efficient identification of the commonly-used foreign language statements and the corresponding scenes; and on the basis, more optimized translation results can be provided.

A corpus generation device and method, the device comprising: a segmentation module, connected to at least one monolingual parallel corpus for segmenting a sentence into words and processing the segmented words by a knowledge-driven approach; a classification module, for classifying sentences having different tag sequences but the same meaning into the same sentence cluster; a mapping module, for determining the categories of sentence structures of all the sentences in the sentence cluster, recording and storing a mapping mode for transforming tags between sentence structures when different categories of sentence structures in the same sentence cluster are transformed; a sentence structure generation module, for generating sentence structures according to a first mapping mode between a first category of sentence structures in one of the sentence clusters and other categories of sentence structures in the same sentence cluster; and a corpus generation module, for nesting a word corresponding to a sequence tag to generate a new monolingual parallel corpus.

The invention discloses a method for realizing support of multiple languages of an application program, a storage device and a mobile terminal. The method comprises the steps that a language identifier base containing language identifiers of all countries is preset in the application program; a language currently used in the mobile terminal is recognized, and the language identifier, corresponding to the language currently used by a user, in the language identifier base is acquired; and a translation system preset in the application program automatically translates a source language on interfaces and functional buttons of the application program into the language, corresponding to the language identifier, currently used in the mobile terminal, and the language is displayed. Through the method, in the development process of the application program, the languages of all the countries do not need to be translated in advance, it is not needed to spend a large amount of capital and time on processing language translation problems, development cost of the application program is saved, translation accuracy is effectively optimized, and convenience is brought to the user.

The invention discloses a translation method, a machine translation model training method and device and a storage medium, and relates to the field of artificial intelligence. The method comprises: obtaining a sample statement marked with a reference translation result; inputting the sample statement into a to-be-trained machine translation model to obtain a sample translation result of the samplestatement, the machine translation model being a non-autoregressive model with to-be-adjusted model parameters; calculating a gradient estimation value of the word translation result according to thereference translation result, and obtaining the gradient estimation value through calculation of a reward value; and adjusting model parameters of the machine translation model according to a gradient estimation result. According to the method, the gradient estimation value is calculated through the reward value, the model parameters of the machine translation model are adjusted according to thegradient estimation value, the gradient estimation value is obtained with the reward value as the adjustment reference data to adjust the model parameters, and therefore sequence-level training of themachine translation model is achieved, and the translation accuracy of the trained machine translation model is improved.

The invention relates to a Chinese-English machine translation method based on context iteration analysis, which comprises the following steps of: (1) judging whether a keyword library, a professionalterm library, a Chinese common incidence word and sentence structure library, a Chinese common phrase library, a word library with context information and a statistical machine translation model exist, if any library or model is missing, entering step (2), otherwise, entering step (3); (2) constructing one keyword library, one professional term library, one Chinese common incidence word and sentence structure library, one Chinese common phrase library, one word library with context information and one statistical machine translation model; (3) receiving a Chinese abstract of a paper to be translated and performing context iterative translation on the Chinese abstract; (4) structuring the English abstract; (5) outputting the English abstract. The Chinese-English machine translation methodbased on context iteration analysis has the advantages of wide translation field, high accuracy and the like.

The invention belongs to the technical field of computer software and discloses a method and a system for segmenting a sentence in machine translation, a computer program and a computer. The method comprises the steps of providing a language category of an input text; carrying out sentence segmentation of the text by use of a sentence segmenting method. With the method and the system for segmenting the sentence in machine translation, the computer program and the computer, the translation quality of a long sentence or a chapter text can be improved, the quality of a translated text obtained bysentence segmentation, separate translation and translation combination is obviously better than the quality of the translated text which is not subjected to sentence segmentation; the problem of long translation time of an ultra-long sentence can be also solved; and the effect of the translation response speed is improved by sentence segmentation and distribution of translation requests. Aimingat the problem of the possible sentence segmentation in the practical application of the machine translation, the invention provides a sentence segmentation method for the machine translation. In a practical application test, the accuracy of sentence translation on a BLEU (Bilingual Evaluation Understudy) is improved by about 0.8 on average and the translation efficiency is improved by about 7%.

The invention relates to the technical field of text translation, and discloses a machine translation algorithm and device based on layer aggregation. The algorithm comprises the steps: obtaining a to-be-translated Chinese statement, and carrying out the preprocessing of the to-be-translated Chinese statement; the ATransformer encoder performs multilayer semantic feature information extraction onthe preprocessed statement on the basis of a multilayer information extraction algorithm; the ATransformer decoder is used for decoding the multi-layer semantic feature information and outputting a translation target language sequence; the judgment model D judges the translation target language sequence; if the translation target language sequence is judged to be a translation result; and if yes,taking the translation target language sequence as a final machine translation result and outputting the final machine translation result. Otherwise, parameter updating is conducted on the ATransformer model based on the strategy gradient algorithm, the preprocessed statement to be translated is input into the updated ATransformer encoder, and the machine translation algorithm is achieved again. The invention further provides a device of the machine translation algorithm based on layer aggregation. According to the invention, intelligent translation of the text is realized.

The invention discloses a text translation method, a related device, electronic equipment and a storage medium, and the text translation method comprises the following steps: obtaining a to-be-translated text and a plurality of reference texts, wherein the to-be-translated text is expressed in a source language, the to-be-translated text contains polysemous words, the plurality of reference texts are expressed in a target language, and each reference text contains paraphrasing words of the polysemous words; extracting a first semantic representation of polysemy words in the to-be-translated text, and respectively extracting a second semantic representation of paraphrasing words in each reference text; based on a clustering result of the first semantic representation and the second semantic representation, obtaining translated words, paraphrased by the target language, of the polysemous words in the to-be-translated text; and translating the to-be-translated text based on the translation words to obtain a translated text represented by the target language. According to the scheme, the translation accuracy can be improved.

The invention discloses a statement segmentation method and device, a storage medium, a processor and terminal equipment. The method comprises the steps of acquiring training data , and the training data are bilingual sentence pairs to be used formed by segmenting initial bilingual sentence pairs at least based on a word alignment relation; training through the training data to obtain a statementsegmentation model; and segmenting the to-be-segmented statement by adopting the statement segmentation model. According to the method and the apparatus, the technical problem of very large translation deviation caused by incapability of effectively segmenting statements and/or sentence pairs during text translation in related technologies is solved.

The invention discloses a translation model training method, a translation method and a translation system for a sign language video in a specific scene, and belongs to the field of video natural language generation. According to the invention, during training, a filtering network is trained through the constructed actor-double-critic deep reinforcement learning training architecture, then the keyframe sequence set of the sign language video is screened out through the trained filtering network, and finally the deep sequence self-encoding network is trained based on deep learning. The filtering network removes the space-time redundancy of the video frames, and the depth sequence self-coding network fuses the attention mechanism, so that the trained sign language video translation model has high translation accuracy, and the translation method and system of the sign language video in the specific scene are further provided. And on the basis that the trained translation model has relatively high translation accuracy, sign language video translation is carried out, so that the accuracy of sign language video translation is effectively improved.