Translation method, device and computer readable storage medium

Abstract

The present invention discloses a translation method, device and computer-readable storage medium, wherein the translation method comprises: determining a first positional relationship between a source text and a style tag in a source file; extracting the source text from the source file according to the first positional relationship; determining a target text corresponding to the source text; and generating a target file corresponding to the source file according to the target text, the style tag and the first positional relationship. The present invention aims to improve the integrity of machine translation.

Description

Technical Field

[0001] The present invention relates to the field of machine translation, and in particular to a translation method, device and computer-readable storage medium. Background Art

[0002] With the rapid development of economic globalization and the Internet, machine translation technology plays an increasingly important role in promoting political, economic and cultural exchanges. Machine translation is the process of using computers to convert one natural language (source language) into another natural language (target language).

[0003] In the process of machine translation of source files, text and style tags are usually machine translated together. Style tags are identifiers for style conversion of text content. Style tags will cut off sentences, which will affect the results of machine translation during the translation process, resulting in low integrity of machine translation.

[0004] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of the present invention is to provide a translation method, device and computer-readable storage medium, aiming to achieve the effect of improving the integrity of machine translation.

[0006] To achieve the above object, the present invention provides a translation method, which comprises the following steps:

[0007] Determine a first position relationship between source text and a style tag in a source file;

[0008] Extracting the source-end text from the source-end file according to the first position relationship;

[0009] Determine a target-side text corresponding to the source-side text;

[0010] A target-side file corresponding to the source-side file is generated according to the target-side text, the style tag, and the first positional relationship.

[0011] Optionally, the step of generating a target-side file corresponding to the source-side file according to the target-side text, the style tag and the first position relationship includes:

[0012] Determining a second positional relationship between the target-end text and the middle-style tag according to the word alignment result and the first positional relationship;

[0013] Determining, according to the second position relationship, an insertion position of the style tag in the target-side text;

[0014] The style tag is inserted into the target-side text according to the insertion position to generate a target-side file.

[0015] Optionally, the step of determining a first positional relationship between the source text and the style tag in the source file includes:

[0016] Determine a text style change node in the source file;

[0017] The first position relationship is determined according to the node position of the changed node.

[0018] Optionally, the step of extracting the source text from the source file according to the first position relationship includes:

[0019] Determine the functional characteristics of the style tag according to the text style of the source text before the node position or corresponding to the node position;

[0020] The style tag is removed from the source file, and the style tag, the functional feature and the first position relationship are associated and saved;

[0021] The source-end file after the style tag is removed is retained as the source-end text.

[0022] Optionally, the step of generating a target-side file corresponding to the source-side file according to the target-side text, the style tag and the first position relationship includes:

[0023] inserting a style tag associated with the first positional relationship into the target-end text according to the first positional relationship;

[0024] The text style of the target-side text is set according to the functional features associated with the style tag to generate the target-side file.

[0025] Optionally, the step of determining the target-end text corresponding to the source-end text includes:

[0026] Determining first content and second content of the source text, wherein the first content and the second content do not overlap;

[0027] Translating the first content to generate a first part of target-end text;

[0028] matching a bilingual corpus associated with the first content;

[0029] The second content is translated according to the bilingual corpus to generate a second part of target-end text.

[0030] Optionally, the step of extracting the source text from the source file according to the first position relationship includes:

[0031] Determine a third position relationship between source text and special symbols in a source file;

[0032] The special symbol is removed from the source file, and the special symbol is associated with the third position relationship and saved.

[0033] Optionally, the step of generating a target-side file corresponding to the source-side file according to the target-side text, the style tag and the first position relationship further includes:

[0034] Inserting the style tag into the target end text according to the first position relationship;

[0035] The special symbol is inserted into the target-side text according to the third positional relationship to generate the target-side file.

[0036] In addition, to achieve the above-mentioned purpose, the present invention also provides a translation device, which includes a memory, a processor, and a translation program stored in the memory and executable on the processor, and the translation program implements the steps of the translation method described above when executed by the processor.

[0037] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, on which a translation program is stored, and when the translation program is executed by a processor, the steps of the translation method described above are implemented.

[0038] A translation method, device and computer-readable storage medium proposed in an embodiment of the present invention first determine a first positional relationship between a source text and a style tag in a source file; extract the source text from the source file based on the first positional relationship; determine a target text corresponding to the source text; and generate a target file corresponding to the source file based on the target text, the style tag and the first positional relationship. When translating a source file, the source text and the style tag of the source file are separated, and after the source text is extracted for translation to generate the target text, the target file is generated based on the first positional relationship. In this way, the style tag will not interrupt the coherence of the sentences in the source file during translation, thereby improving the integrity of machine translation. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a schematic diagram of the terminal structure of the hardware operating environment involved in the embodiment of the present invention;

[0040] Figure 2A schematic diagram of a flow chart of an embodiment of a translation method of the present invention;

[0041] Figure 3 FIG. 4 is a flow chart of another embodiment of the translation method of the present invention.

[0042] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0043] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0044] In the related art, the text and style tags of the source file are usually machine translated together. The style tags are identifiers for style conversion of the text content. The style tags will cut off the sentences and affect the results of the machine translation during the translation process, resulting in low integrity of the machine translation.

[0045] In order to improve the integrity of machine translation, the embodiment of the present invention provides a translation method, device and computer-readable storage medium, wherein the main steps of the method include:

[0046] Determine a first position relationship between source text and a style tag in a source file;

[0047] Extracting the source-end text from the source-end file according to the first position relationship;

[0048] Determine a target-side text corresponding to the source-side text;

[0049] A target-side file corresponding to the source-side file is generated according to the target-side text, the style tag, and the first positional relationship.

[0050] In this way, when translating the source file, the source text is extracted for translation based on the first position relationship between the source text and the style tag. After the target text is generated, the target file is generated based on the first position relationship. In this way, the style tag will not interrupt the coherence of the sentences in the source file during translation, thereby improving the integrity of the machine translation.

[0051] The content to be protected by the claims of the present invention will be described in detail below with reference to the accompanying drawings.

[0052] like Figure 1 As shown, Figure 1 It is a schematic diagram of the terminal structure of the hardware operating environment involved in the embodiment of the present invention.

[0053] The terminal in the embodiment of the present invention may be a translation device.

[0054] like Figure 1As shown, the terminal may include: a processor 1001, such as a CPU, a memory 1003, and a communication bus 1002. The communication bus 1002 is used to realize the connection and communication between these components. The memory 1003 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1003 may also be a storage device independent of the aforementioned processor 1001.

[0055] Those skilled in the art will understand that Figure 1 The terminal structure shown in the figure does not constitute a limitation on the terminal, and may include more or less components than shown in the figure, or combine certain components, or arrange the components differently.

[0056] like Figure 1 As shown, the memory 1003 as a computer storage medium may include an operating system and a translation program.

[0057] exist Figure 1 In the terminal shown, the processor 1001 can be used to call the translation program stored in the memory 1003 and perform the following operations:

[0058] Determine a first position relationship between source text and a style tag in a source file;

[0059] Extracting the source-end text from the source-end file according to the first position relationship;

[0060] Determine a target-side text corresponding to the source-side text;

[0061] A target-side file corresponding to the source-side file is generated according to the target-side text, the style tag, and the first positional relationship.

[0062] Further, the processor 1001 may call the translation program stored in the memory 1003, and further perform the following operations:

[0063] Determining a second positional relationship between the target-end text and the middle-style tag according to the word alignment result and the first positional relationship;

[0064] Determining, according to the second position relationship, an insertion position of the style tag in the target-side text;

[0065] The style tag is inserted into the target-side text according to the insertion position to generate a target-side file.

[0066] Further, the processor 1001 may call the translation program stored in the memory 1003, and further perform the following operations:

[0067] Determine a text style change node in the source file;

[0068] The first position relationship is determined according to the node position of the changed node.

[0069] Further, the processor 1001 may call the translation program stored in the memory 1003, and further perform the following operations:

[0070] Determining the functional characteristics of the style tag according to the text style of the source text corresponding to the node position;

[0071] The style tag is removed from the source file, and the style tag, the functional feature and the first position relationship are associated and saved;

[0072] The source-end file after the style tag is removed is retained as the source-end text.

[0073] Further, the processor 1001 may call the translation program stored in the memory 1003, and further perform the following operations:

[0074] inserting the style tag associated with the first positional relationship into the target end text according to the first positional relationship;

[0075] The text style of the target-side text is set according to the functional features associated with the style tag to generate the target-side file.

[0076] Further, the processor 1001 may call the translation program stored in the memory 1003, and further perform the following operations:

[0077] Determining first content and second content of the source text, wherein the first content and the second content do not overlap;

[0078] Translating the first content to generate a first part of target-end text;

[0079] matching a bilingual corpus associated with the first content;

[0080] The second content is translated according to the bilingual corpus to generate a second part of target-end text.

[0081] Further, the processor 1001 may call the translation program stored in the memory 1003, and further perform the following operations:

[0082] Determining a third position relationship between the source text and the special symbol;

[0083] The special symbol is removed from the source text, and the special symbol is associated with the third positional relationship and saved.

[0084] Further, the processor 1001 may call the translation program stored in the memory 1003, and further perform the following operations:

[0085] Inserting the style tag into the target end text according to the first position relationship;

[0086] The special symbol is inserted into the target-side text according to the third positional relationship to generate the target-side file.

[0087] With the rapid development of economic globalization and the Internet, machine translation technology plays an increasingly important role in promoting political, economic and cultural exchanges. Machine translation is the process of using computers to convert one natural language (source language) into another natural language (target language).

[0088] In the process of machine translation of source files, text and style tags are usually machine translated together. Style tags are identifiers for style conversion of text content. Style tags will cut sentences, resulting in incoherent text content, which will affect the results of machine translation during the translation process, resulting in low integrity of machine translation. Some style tags also contain text content, which will affect the results of machine translation, resulting in low accuracy of machine translation.

[0089] Therefore, the above defects exist in the related translation methods. To solve the above defects, the embodiment of the present invention proposes a translation method, which aims to achieve the first position relationship between the source text and the style tag, extract the source text from the source file for translation, so as to improve the accuracy of translation.

[0090] The following is an explanation of the content claimed in the claims of the present invention through specific exemplary schemes, so that those skilled in the art can better understand the scope of protection of the claims of the present invention. It is understood that the following exemplary schemes do not limit the scope of protection of the present invention, but are only used to explain the present invention.

[0091] For example, refer to Figure 2 In one embodiment of the translation method of the present invention, the translation method comprises the following steps:

[0092] Step S10, determining a first position relationship between the source text and the style tag in the source file;

[0093] In this embodiment, the source file is the original text that needs to be translated. The source file is input into the machine translation model to generate the corresponding translation and the corresponding target file. The style tag is a machine translation plug-in inserted into the source file. The style tag is an identifier for styling and converting the text content. The style tag can be located at the front end, back end or middle end of the sentence, and sets the font, color, font size and other attributes of the corresponding text part. It is a tag with functional characteristics.

[0094] Style tags may be displayed in the source file as symbols without literal meaning, such as *, #, etc. Such style tags can cut off sentences and can be ignored during translation. They have little effect on the translation results at the beginning or end of a sentence, but they will still cut off sentences in the middle of a sentence, affecting the integrity of the sentence translation. Style tags may also be symbols with literal meaning, such as (Kaiti, No. 4), etc. Such style tags will not only cut off sentences and affect the integrity of sentence translation, but the machine translation model will also translate the text content of the style tags, affecting the accuracy of sentence translation.

[0095] The source file includes source text and style tags. The source text includes the main content of the source file. The style tag is inserted into the source text content. The first position relationship between the style tag and the source text can be determined according to the position of the style tag in the source text.

[0096] Optionally, a text style change node in the source file is determined, and the first positional relationship is determined according to the node position of the change node. The insertion of a style tag will change the text style of the source text. If a text style change node appears in the source file, and the text style before the change node is different from the text style after the change node, then the change node has a style tag, that is, the node position of the change node can determine the position of the style tag, and the first positional relationship can be determined according to the position of the style tag.

[0097] Step S20: extracting the source text from the source file according to the first position relationship.

[0098] In this embodiment, the first positional relationship is the positional relationship between the style tag and the source text. The style tag can be indexed based on the first positional relationship. The style tag can be removed from the source text to generate the source text. The source text can be plain text content, including basic punctuation marks.

[0099] Optionally, the functional characteristics of the style tag are determined according to the text style of the source text corresponding to the node position; the style tag is removed from the source file, and the style tag, the functional characteristics and the first position relationship are associated and saved; and the source file after the style tag is removed is retained as the source text.

[0100] When the style tag is removed from the source file, the functional characteristics of the style tag will not affect the source text. However, in order to improve the comparability between the target file and the source file, the style tag can be inserted back into the target text when the target file is generated so that the style of the generated target file can correspond to the style of the source file. Therefore, the functional characteristics of the style tag need to be retained.

[0101] First, the functional characteristics of the style tag are determined according to the text style of the paragraph text corresponding to the node position of the style tag. The text style of the source text before or after the node position can be determined according to preset rules to determine the functional characteristics of the style tag. For example, if the text style of the source text before the style tag can be changed after the style tag is inserted into the text, then the text style before the node position is determined to be the text style corresponding to the style tag. Secondly, after determining the functional characteristics of the style tag, remove the style tag from the source file, and save the style tag, functional characteristics and first position relationship in association. Finally, a source file without a style tag is generated, and the source text can be completed.

[0102] Step S30, determining the target end text corresponding to the source end text;

[0103] In this embodiment, the source text does not include style tags. The source text can be input into a machine translation model and a target text can be output. That is, the source text is translated by a machine translation model to generate a target text. The target text has the same semantics as the source text, but belongs to a different language and does not have a text tag.

[0104] Optionally, a first content and a second content of the source text are determined, wherein the first content and the second content do not overlap; the first content is translated to generate a first part of the target text; a bilingual corpus associated with the first content is matched; and the second content is translated according to the bilingual corpus to generate a second part of the target text.

[0105] In the process of translating the source text by the machine translation model, the source text can be divided into multiple contents, and the first content and the second content are determined, and the second content is the adjacent content after the first content. When the machine translation model translates the source text, the first content is translated first to generate the corresponding first part of the label end file, and then more related bilingual corpora are matched according to the first content, and the bilingual corpora in the same professional field associated with the first content are matched. Specifically, the bilingual corpora in the same technical field as the sentences of the first content can be matched in the memory library, and the bilingual corpora in the same technical field as the vocabulary of the first content can be matched in the terminology library. The machine translation model is incrementally trained based on the bilingual corpora to improve the accuracy of the machine translation model in translating the text in the professional field. It is also possible to index to the database associated with the first content according to the preset corresponding relationship, and perform incremental training on the machine single model according to the corpus data in the database. In this way, after the first content is translated, the second content is translated based on the incrementally trained machine translation model to generate the second part of the target end text, and the above steps are repeated until the source end text is fully translated. During the translation process, the machine translation model continuously improves its own translation accuracy, and the accuracy of the translated source text is higher.

[0106] Step S40: Generate a target-side file corresponding to the source-side file according to the target-side text, the style tag, and the first position relationship.

[0107] In this embodiment, after obtaining the target-side text, a target-side file corresponding to the source-side file is generated according to the target-side text, the style tag and the first position relationship. The position of the style tag in the target-side file corresponds to the position in the source-side file. Therefore, the text style of the target-side file corresponds to the text style of the source-side file. When translating the source-side file, the parts with the same content in the source-side file and the target-side file can be found according to the same text style, thereby improving the comparability of the original text and the translated text.

[0108] Optionally, based on the word alignment result and the first positional relationship, a second positional relationship between the target-side text and the style tag is determined; based on the second positional relationship, an insertion position of the style tag in the target-side text is determined; and based on the insertion position, the style tag is inserted into the target-side text to generate a target-side file.

[0109] The word alignment technology is used to align the words, sentences or paragraphs in the source text and the target text to obtain the comparison relationship between the source text and the target text content, which is the word alignment result. According to the word alignment result, the source text in the first position relationship is replaced with the target text, and the second position relationship between the style tag and the target text is obtained. According to the second position relationship, the style tag removed from the source file before translation is inserted back into the target text. Specifically, the insertion position of the style tag in the target text is determined according to the second position relationship, and the style tag is inserted into the insertion position of the target text to generate the target file. In this way, the generated target file has a higher comparison with the source file, and the integrity of the translated source file is better.

[0110] In the technical solution disclosed in this embodiment, the first positional relationship between the source text and the style tag in the source file is first determined; according to the first positional relationship, the source text is extracted from the source file; the target text corresponding to the source text is determined; according to the target text, the style tag and the first positional relationship, a target file corresponding to the source file is generated. In this way, when translating the source file, the source text and the style tag of the source file are separated, and after the source text is extracted for translation to generate the target text, the target file is generated according to the first positional relationship. There is no style tag during translation, and the coherence of the sentences in the source file will not be interrupted. After the translation is completed, the target text is generated according to the target text and the first positional relationship. The target text has a better comparison with the source text, and the text style should not be changed except for changing the language, thereby improving the integrity of machine translation.

[0111] Optionally, refer to Figure 3 Based on any of the above embodiments, in another embodiment of the translation method of the present invention, the translation method further includes:

[0112] Step S40, determining a third position relationship between the source text and the special symbol;

[0113] Step S50: remove the special symbol from the source file, and store the special symbol in association with the third position relationship.

[0114] In this embodiment, the source text contains special symbols in addition to style tags. Special symbols have no functionality in the source file and only play a role in the content connection of the source text. They have no actual meaning, but like style tags, they will cut off sentences. Most special symbols cannot be recognized by the machine translation model. When translating, special symbols in a sentence are mistakenly regarded as the segmentation point of the sentence, and a complete sentence is misjudged as two sentences. For example, the corner marks commonly seen in papers are easily misjudged.

[0115] After the style tags are removed from the source file, pre-set special symbols that need to be removed or special symbols that the machine translation model has not been trained to recognize and cannot recognize can be identified by means of image recognition or the like, and the position of the special symbols in the source file can be determined, and then the third position relationship between the source text and the special symbols can be determined. The special symbols are removed from the source text based on the third position relationship, and the special symbols are associated with the third position relationship and saved.

[0116] Optionally, step S40 includes: inserting the style tag into the target-side text according to the first positional relationship; inserting the special symbol into the target-side text according to the third positional relationship to generate the target-side file.

[0117] The style tags are removed from the source file according to the first position relationship to obtain the source text, and the special symbols are extracted from the source text according to the third position relationship to obtain a text file that is easier to be recognized by the machine translation model. In order to ensure the integrity of the translated source file, the target file can retain the style tags and special symbols corresponding to the source file. Therefore, the insertion position of the style tags in the target text can be determined according to the word alignment result and the first position relationship, and the style tags are inserted into the target text. The insertion position of the special symbols in the target text can be determined according to the word alignment result and the third position relationship, and the special symbols are inserted into the target text. Finally, the text style and content of the target text obtained can correspond to the source file.

[0118] In the technical solution disclosed in this embodiment, the third position relationship between the source text and the special symbol is determined; the special symbol is removed from the source text, and the special symbol and the third position relationship are associated and saved. This can prevent the special symbol from cutting off the sentence during translation, thereby improving the completeness and accuracy of machine translation. After the translation is completed, the special symbol and the style tag are correspondingly inserted into the target text, which can also improve the comparison between the source file and the target file.

[0119] In addition, an embodiment of the present invention also proposes a translation device, which includes a memory, a processor, and a translation program stored in the memory and executable on the processor, wherein the translation program implements the steps of the translation method described in the above embodiments when executed by the processor.

[0120] In addition, an embodiment of the present invention further provides a computer-readable storage medium, on which a translation program is stored. When the translation program is executed by a processor, the steps of the translation method described in the above embodiments are implemented.

[0121] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0122] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0123] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes a number of instructions for enabling the translation device to execute the methods described in each embodiment of the present invention.

[0124] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A translation method, characterized in that: The translation method comprises the following steps: Determine a first positional relationship between a source text and a style tag in a source file, wherein a text style change node in the source file is determined, and the first positional relationship is determined according to a node position of the change node; Determining the functional characteristics of the style tag according to the text style of the source text corresponding to the node position; The style tag is removed from the source file, and the style tag, the functional feature and the first position relationship are associated and saved; retaining the source file after removing the style tag as the source text; Determine a target text corresponding to the source text, wherein a first content and a second content of the source text are determined, the first content and the second content do not overlap, the first content is translated to generate a first part of the target text, a bilingual corpus associated with the first content is matched, and the second content is translated according to the bilingual corpus to generate a second part of the target text; Determining a second positional relationship between the target-end text and the style tag according to the word alignment result and the first positional relationship; Determining, according to the second position relationship, an insertion position of the style tag in the target-side text; The style tag is inserted into the target-side text according to the insertion position to generate a target-side file.

2. The translation method according to claim 1, characterized in that: The step of generating a target-side file corresponding to the source-side file according to the target-side text, the style tag and the first position relationship comprises: inserting a style tag associated with the first positional relationship into the target-end text according to the first positional relationship; The text style of the target-side text is set according to the functional features associated with the style tag to generate the target-side file.

3. The translation method according to claim 1, characterized in that: After the step of extracting the source text from the source file according to the first position relationship, the method further includes: Determining a third position relationship between the source text and the special symbol; The special symbol is removed from the source text, and the special symbol is associated with the third positional relationship and saved.

4. The translation method according to claim 1, characterized in that: The step of generating a target-side file corresponding to the source-side file according to the target-side text, the style tag and the first position relationship comprises: Inserting the style tag into the target end text according to the first position relationship; The special symbol is inserted into the target-side text according to the third positional relationship to generate the target-side file.

5. A translation device, characterized in that: The translation device comprises: a memory, a processor, and a translation program stored in the memory and executable on the processor. When the translation program is executed by the processor, the steps of the translation method according to any one of claims 1 to 4 are implemented.

6. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a translation program, and when the translation program is executed by a processor, the steps of the translation method according to any one of claims 1 to 4 are implemented.

Images