阅读说明：本技术 翻译方法、装置、设备及存储介质 (Translation method, device, equipment and storage medium ) 是由 林原 于 2019-10-10 设计创作，主要内容包括：本发明公开了一种翻译方法、装置、设备及存储介质,所述翻译方法包括步骤：识别出待翻译成目标语言的原文存在多种意义；根据问答的方式确定所述原文所要表达的单一意义；根据确定的单一意义,将构成所述原文的源语言转化成所述目标语言。本发明的问答过程完全使用源语言,可以使得用户不需要精通目标语言的情况下,提高计算机辅助翻译的翻译质量。(The invention discloses a translation method, a translation device, translation equipment and a storage medium, wherein the translation method comprises the following steps: recognizing that the original text to be translated into the target language has multiple meanings; determining the single meaning of the original text to be expressed according to a question-answering mode; and converting the source language forming the original text into the target language according to the determined single meaning. The question-answering process of the invention completely uses the source language, so that the translation quality of computer-aided translation can be improved under the condition that the user does not need to be proficient in the target language.)

1. A translation method, characterized in that the translation method comprises the steps of:

recognizing that the original text to be translated into the target language has multiple meanings;

determining the single meaning of the original text to be expressed according to a question-answering mode;

and converting the source language forming the original text into the target language according to the determined single meaning.

2. The translation method according to claim 1, wherein said recognizing that the original text to be translated into the target language has a plurality of meanings comprises the steps of:

when words with multiple meaning words and/or multiple parts of speech in the original text are identified, and/or when grammatical structure errors or ambiguities of the original text are identified, multiple meanings of the original text are identified.

3. The translation method according to claim 1 or 2, wherein said determining a single meaning to be expressed by said original text in a question-and-answer manner comprises the steps of:

proposing a question using the source language;

receiving an answer described in a source language in accordance with a proposed question;

determining a single meaning to be expressed by the original text based on the received answer.

4. The translation method of claim 3, wherein said posing a question in said source language comprises:

analyzing the cause of the multiple meanings;

determining the problem according to the reason;

the problem is posed using the source language.

5. The translation method of claim 3, wherein said receiving an answer described in a source language based on a proposed question comprises one or more of:

receiving a selected single meaning from a plurality of meanings described in the source language;

confirming the single meaning of the original text by adopting the source language prompt;

and modifying the original text by adopting the source language prompt.

6. A translation method according to claim 3, wherein said receiving of an answer described in a source language in response to a proposed question, comprises the steps of:

taking the received answer as an answer corresponding to the question to be provided, and storing the answer in an answer database;

determining the answer preferred by the user according to the received answer, taking the preferred answer as the answer corresponding to the question and storing the answer in the user preference database;

and carrying out statistical analysis and/or machine learning on the received answers, obtaining machine answers through the statistical analysis and/or the machine learning as answers corresponding to the questions, and storing the machine answers into a statistical analysis and machine learning database.

7. The translation method according to claim 6, wherein said converting a source language constituting said original text into said target language according to the determined univocal meaning, comprises the steps of:

searching answers corresponding to the questions from the answer database, the user preference database and the statistical analysis and machine learning database;

when the answer is found, extracting the answer, and determining the single meaning to be expressed by the original text according to the extracted answer;

and when the original text is not found, determining the single meaning to be expressed by the original text according to a question-answering mode.

8. A translation apparatus, characterized in that the translation apparatus comprises:

the recognition module is used for recognizing that the original text to be translated into the target language has multiple meanings;

the questioning module is used for determining the single meaning to be expressed by the original text according to a questioning and answering mode;

and the translation generation module is used for converting the source language forming the original text into the target language according to the determined single meaning.

9. A translation apparatus, characterized in that the translation apparatus comprises: a memory and a processor, the memory storing a translation program which, when executed by the processor, implements the steps of the method of any of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a translation program executable by at least one processor to perform the steps of the method of any one of claims 1 to 7.

Technical Field

The present invention relates to the field of computer translation, and in particular, to a translation method, apparatus, device, and storage medium.

Background

Computer-assisted translation (CAT) means that a translator translates an original text (source language) into a translated text (target language) with the help and prompting of a Computer. Although computer-aided translation can solve the problem of inaccurate machine translation software, the problems of great difficulty in use, high labor cost and requirement for the user to master the source language and the target language also exist in common users (non-professional translators). For example, since most translators are only skilled in one target language, many translators are required to translate an article into multiple target languages, and especially, there are fewer translators in the small languages, which results in higher labor cost and requires professional translators, and thus, the use difficulty of common users is high.

Based on the above, how to ensure the translation quality and reduce the use difficulty and labor cost of computer-aided translation is not an effective solution in the prior art.

Disclosure of Invention

The embodiment of the invention provides a translation method, a translation device, translation equipment and a storage medium, which are used for improving the translation quality of computer-aided translation.

In a first aspect, an embodiment of the present invention provides a translation method, where the translation method includes:

recognizing that the original text to be translated into the target language has multiple meanings;

determining the single meaning of the original text to be expressed according to a question-answering mode;

and converting the source language forming the original text into the target language according to the determined single meaning.

Optionally, the recognizing that the original text to be translated into the target language has multiple meanings includes the steps of:

when words with multiple meaning words and/or multiple parts of speech in the original text are identified, and/or when grammatical structure errors or ambiguities of the original text are identified, multiple meanings of the original text are identified.

Optionally, the determining the single meaning to be expressed by the original text according to a question-and-answer mode includes the steps of:

proposing a question using the source language;

receiving an answer described in a source language in accordance with a proposed question;

determining a single meaning to be expressed by the original text based on the received answer.

Optionally, the presenting a question using the source language includes:

analyzing the cause of the multiple meanings;

determining the problem according to the reason;

the problem is posed using the source language.

Optionally, the receiving an answer described in the source language according to the proposed question includes one or more of:

receiving a selected single meaning from a plurality of meanings described in the source language;

confirming the single meaning of the original text by adopting the source language prompt;

and modifying the original text by adopting the source language prompt.

Optionally, after receiving the answer described in the source language according to the proposed question, the method includes the steps of:

taking the received answer as an answer corresponding to the question to be provided, and storing the answer in an answer database;

determining the answer preferred by the user according to the received answer, taking the preferred answer as the answer corresponding to the question and storing the answer in the user preference database;

and carrying out statistical analysis and/or machine learning on the received answers, obtaining machine answers through the statistical analysis and/or the machine learning as answers corresponding to the questions, and storing the machine answers into a statistical analysis and machine learning database.

Optionally, before the converting the source language constituting the original text into the target language according to the determined single meaning, the method includes the steps of:

searching answers corresponding to the questions from the answer database, the user preference database and the statistical analysis and machine learning database;

extracting answers when the answers are found, and determining the single meaning to be expressed by the original text according to the extracted answers;

and when the original text is not found, determining the single meaning to be expressed by the original text according to a question-answering mode.

In a second aspect, an embodiment of the present invention provides a translation apparatus, where the translation apparatus includes:

the recognition module is used for recognizing that the original text to be translated into the target language has multiple meanings;

the questioning module is used for determining the single meaning to be expressed by the original text according to a questioning and answering mode;

and the translation generation module is used for converting the source language forming the original text into the target language according to the determined single meaning.

In a third aspect, an embodiment of the present invention provides a translation apparatus, where the translation apparatus includes: a memory storing a translation program which, when executed by the processor, implements the steps of the method of any one of the above.

In a fourth aspect, embodiments of the present invention provide a computer-readable storage medium having a translation program stored thereon, the translation program being executable by at least one processor to implement the steps of the method as described in any one of the above.

The embodiment of the invention identifies the multiple meanings of the original text to be translated into the target language and determines the single meaning to be expressed by the original text according to the question-answering mode, so that the source language forming the original text can be converted into the target language according to the determined single meaning, the meaning to be really expressed by a user is identified before computer-assisted translation, the translation process is controlled, and the translation quality of the computer-assisted translation is improved under the condition that the user does not need to master the target language.

The foregoing description is only an overview of the technical solutions of the present invention, and the embodiments of the present invention are described below in order to make the technical means of the present invention more clearly understood and to make the above and other objects, features, and advantages of the present invention more clearly understandable.

Drawings

Various other advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:

FIG. 1 is a flow chart of a translation method according to an embodiment of the present invention;

FIG. 2 is a diagram of a conventional computer-aided translation of corrected terms;

FIG. 3 is a diagram illustrating computer-aided translation of corrected terms, in accordance with an embodiment of the present invention;

fig. 4 is a schematic structural diagram of a translation apparatus according to an embodiment of the present invention.

Detailed Description

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

In the following description, suffixes such as "module", "component", or "unit" used to denote elements are used only for facilitating the explanation of the present invention, and have no specific meaning in itself. Thus, "module", "component" or "unit" may be used mixedly.

A first embodiment of the present invention provides a method for translating a source language (language of a source text), as shown in fig. 1, including the following steps:

step S101, recognizing that the original text to be translated into the target language has multiple meanings;

step S102, determining the single meaning to be expressed by the original text according to a question-answering mode;

and step S103, converting the source language forming the original text into the target language according to the determined single meaning.

The embodiment of the invention can be executed in a computer, a mobile terminal, intelligent wearable equipment, a tablet personal computer and translation equipment.

Where the textual representation is a word, article, sentence, or the like that the user is to translate. The source language indicates a language composition of the original text, for example, an original text composed of chinese, which is a source language belonging to the original text. The target language represents the language to which the original text needs to be translated, e.g., chinese needs to be translated to english, which then belongs to the target language. The target language may be multiple simultaneously in embodiments of the present invention.

In the specific implementation process of the embodiment of the invention, each sentence in the original text can be translated through the steps. In step S101, the grammar structure and word meaning of the original text are analyzed to determine whether the original text has multiple meanings. The parsing syntax structure is to find sentence components such as subject, predicate, object, subject, and final phrase of a sentence, whether there are errors and ambiguities in the sentence structure, and the collocation relationship between words (e.g., a moving object relationship). Analyzing the meaning of a word means to distinguish the part of speech (e.g., whether it is a noun or a verb) of a word, whether the word is an ambiguous word, etc. In the specific implementation process, the parser can be used for analyzing the syntactic structure and word meaning of the original text, and if the parsing result of the parser is multiple, the original text is indicated to have multiple meanings. The algorithm of the resolver can be performed by a person skilled in the art, and is not described in detail herein.

In step S102, a question may be posed for a grammar structure error, ambiguity, ambiguous word, etc., so as to obtain an answer to be answered by the user, and then a single meaning to be actually expressed in the original text is determined according to the answer. Further, in step S103, the original text may be translated into a translation composed of the target language based on the determined univocal meaning and the analysis result in S101.

According to the embodiment of the invention, the original text to be translated into the target language has multiple meanings, and the single meaning to be expressed by the original text is determined according to the question and answer mode, so that the source language forming the original text can be converted into the target language according to the determined single meaning, the meaning to be really expressed by a user is identified before computer translation, the translation process is controlled, and the translation quality of computer-aided translation is improved under the condition that the user does not need to master the translated text.

For example, after the existing computer-aided translation software translates the original text into the translated text, the user finds out the error in the reference translated text compared with the original text, and at this time, the user needs to be proficient in the target language of the translated text.

On the basis of the above embodiments, several specific and alternative embodiments are given below to refine and optimize the embodiments of the present invention, so that the implementation of the scheme of the embodiments of the present invention is more convenient and accurate. In addition, the following embodiments may be arbitrarily combined with each other without conflict.

In some embodiments, optionally, the recognizing that the original text to be translated into the target language has multiple meanings includes the steps of:

when words with multiple meaning words and/or multiple parts of speech in the original text are identified, and/or when grammatical structure errors or ambiguities of the original text are identified, multiple meanings of the original text are identified.

In a specific implementation process, a grammatical structure error or ambiguity of the original text can be identified according to a preset grammatical structure of the source language

In some embodiments, optionally, the determining the single meaning to be expressed by the original text according to a question-and-answer mode includes the steps of:

proposing a question using the source language; receiving an answer described in a source language according to a proposed question; determining a single meaning to be expressed by the original text based on the received answer.

In the embodiment, the question is asked in the source language, so that the question answering process is carried out by adopting the language of the original text, which is the native language of the original text author, and the original text author and the industry expert can control the translated text, thereby effectively reducing the labor cost of the embodiment of the invention, enabling a user to obtain high-quality translation through computer-aided translation only by the native language, effectively improving the universality of the computer-aided translation, and reducing the use difficulty of the original text author, the industry expert and the like which are not good at the target language in the computer-aided translation.

The using the source language to pose a problem includes: analyzing the cause of the multiple meanings; determining the problem according to the reason; the problem is posed using the source language.

For example, if the grammar structure is wrong, a question is posed for the user to correct the grammar structure; if the word is a polysemous word, a question is posed to let the user select one of the word senses.

In some embodiments, optionally, receiving the answer described in the source language according to the proposed question includes one or more of:

the first method is as follows: receiving a selected single meaning from a plurality of meanings described in the source language;

the second method comprises the following steps: confirming the single meaning of the original text by adopting the source language prompt;

the third method comprises the following steps: and modifying the original text by adopting the source language prompt.

That is, in order to allow a user with a limited target language level to also use computer-assisted translation, in some embodiments not only questions may be presented in the source language, but answers may also be received in the source language. Of course, each of the above steps has a suitable range. For example, the question in terms of ambiguous words is most convenient. However, if the structure of the sentence is more cluttered, the third mode is the best choice. Of course, in some embodiments, the questions are of primary and secondary scores, some questions must be answered, and some questions may not be answered. The importance of the question can be distinguished by font color.

In some embodiments, optionally, after receiving the answer described in the source language according to the proposed question, the method includes the steps of:

taking the obtained answer as an answer corresponding to the question to be provided, and storing the answer in an answer database;

determining the answer preferred by the user according to the obtained answer, taking the preferred answer as the answer corresponding to the question and storing the answer in the user preference database;

and carrying out statistical analysis and/or machine learning on the obtained answers, taking the machine answers obtained by the statistical analysis and/or the machine learning as answers corresponding to the questions, and storing the answers in a statistical analysis and machine learning database.

In some embodiments, optionally, before converting the source language constituting the original text into the target language according to the determined single meaning, the method comprises the steps of:

searching answers corresponding to the questions from the answer database, the user preference database and the statistical analysis and machine learning database;

extracting answers when the answers are found, and determining the single meaning to be expressed by the original text according to the extracted answers;

and when the original text is not found, determining the single meaning to be expressed by the original text according to a question-answering mode.

That is, the answer database stores questions and answers that the user has answered; the user preference database stores the user's words and grammatical preferences, e.g., if the user is a machine engineer who is translating an article about the machine, then words about the machine are added to the database; the statistical analysis and machine learning database stores the results of the statistical analysis and machine learning of the question-answering process. For example, without modifiers, how to decide whether a noun is singular or plural. The statistical results are added to the database as default values.

Further, when a question such as a grammar error, an ambiguity, an ambiguous word, etc. is found, step S102 does not present the question, and may search the answer database to ensure that the same question is not repeatedly presented. Searching a user preference database, selecting favorite answers of the user, and if the user translates an article in the aspect of machinery, selecting relevant mechanical meanings of polysemous words. Finally, the statistical analysis and machine learning database is searched to see if default values are available. If the answers cannot be found in the searches, the questions can be provided for the user, the same questions can be guaranteed not to be repeated under most conditions, and therefore the use complexity and difficulty of the computer-aided translation are reduced on the basis of effectively improving the quality of the computer-aided translation, and the efficiency of the question answering process is effectively improved.

The following describes a translation method according to an embodiment of the present invention with a translation example.

The original text "quality we want to pay attention" is translated into english by computer assistance.

First, the grammar structure of the original text and the meaning of the word in the original text are recognized. It can be recognized that the subject is "we", the predicate is "to be noticed", and the object is "quality"; wherein, the relationship between 'our' and 'attention' is a main-meaning relationship, and the relationship between 'attention' and 'quality' is a moving guest relationship; there are no errors and ambiguities in the grammar structure. But it is found that "quality" is an ambiguous word, one is the quality of the product, whether it is durable, etc., and one is a physical quantity, which indicates how much of a substance is. Thus, there are a number of meanings to identify the original. In some embodiments, the words in the original text with the ambiguous words and/or the polysemous may be identified according to a pre-stored dictionary of the ambiguous words and the polysemous.

Secondly, questions are put forward in the source language for grammar, ambiguity, polysemous words and the like, and the single meaning to be expressed by the original text is determined in a question-answering mode. Where the source language may be described as the user's native language.

For example, since "quality" is a multiple meaning word, the meaning of "quality" cannot be determined from the collocation relationship between words. In order to clarify the meaning of "quality", a question-and-answer method is adopted.

Then, the "quality" can be displayed on the preset user interface, and has two meanings, namely quality and physical quantity. The user selects the meaning of "quality" on the user interface. The question and the choice are both languages of the original text, and the user can complete the question and answer only by using the native language of the user. That is, a selected single meaning is received from among a plurality of meanings described in the source language.

Finally, the original text is translated into a translation according to the user's selection. If the user selects "quality", it turns "quality" to "quality", and if the user selects "physical quantity", it turns "quality" to "mass".

The prior computer-aided translation firstly translates an original text into a reference translation, and then a user finds out errors in the reference translation by comparing the original text, so the user has to be proficient in two languages, which cannot be done by most original text authors and industry experts. The question-answering process of the embodiment of the invention is carried out by using the source language of the original text, which is the native language of the original text author, so that the original text author and an industry expert can control the translated text, thereby effectively reducing the difficulty of computer-aided translation, effectively providing the quality of computer-aided translation and effectively reducing the labor cost.

The embodiment of the invention can simultaneously translate the original text into a plurality of target languages, does not need to compare errors of reference translation of the source language and any target language, and can realize the conversion of the source language into the plurality of target languages by solving problems once.

For example, the original text "quality we want to pay attention" is translated into english, spanish and french by computer assistance.

As shown in fig. 2, when the existing computer-aided translation software encounters the ambiguous word "quality", it uses its own algorithm to make a judgment on the word meaning, and then translates the word meaning into a reference translation, such as "quality" in english, "caliadad" in spanish, and "quality" in french. However, when the user performs calibration, the user of the middle translation needs to change "quality" to "mass", the user of the middle translation needs to change "caliddad" to "masa", and the user of the middle translation needs to change "quality" to "mass". Therefore, labor costs are high.

In the embodiment of the invention, the question can be asked once through the source language, so that the specific meaning of the polysemous word to be expressed can be determined, further, the translation can be carried out once, and the labor cost is effectively reduced. As shown in fig. 3, after finding the ambiguous word "quality", the user asks questions in the user's native language, and also answers and selects "physical quantity" in the native language. Then, the software automatically and accurately turns "quality" to "mass" in English, "masa" in Spanish, and "mass" in French. The needed labor cost is counted, and the fact that three users who are proficient in two languages are needed by the existing computer-aided translation software can be found, but only one user needs to carry out question answering once by using the native language of the user, and therefore labor cost and time cost are effectively reduced.

Therefore, in the embodiment of the invention, the user directly controls the translation generated by the computer-aided translation software in a question-and-answer mode; using a source language in the question and answer process; the problem provided by computer-aided translation is that the computer-aided translation software can not make a decision under the existing rule; the user answers in the following ways: a) selecting one from a plurality of choices; b) confirming a selection; c) modifying the original text; computer-aided translation can remember previous questions and answers, and the same questions will not be repeated in most cases; the computer-aided translation can remember the answers of the user and establish a personal preference database of the user; the question answering process is subjected to statistical analysis and machine learning, so that the efficiency of the question answering process is higher.

The embodiment of the invention has the advantage of accurate and controllable translation. For example, in the question-answering process, a translator can correct errors in the original text, eliminate ambiguity, determine the meaning of a polysemous word, and finally generate a correct translation by using the invention. The question-answering process of the invention has a simple and clear user interface, and the user operation is very convenient.

Effectively reduce the labor cost. The embodiment of the invention reduces the labor cost of translation from two aspects. First, professional translation, original text authors and users with limited foreign language level can use the embodiment of the invention to translate, thereby expanding the universality of the embodiment of the invention. Secondly, when the article is translated into various characters, the embodiment of the invention can automatically generate the translations of various target languages only by performing question answering once, and the labor cost is a fraction of the conventional computer-aided translation.

A second embodiment of the present invention provides a translation apparatus, as shown in fig. 4, including:

the recognition module 10 is used for recognizing that the original text to be translated into the target language has multiple meanings;

the questioning module 12 is used for determining the single meaning to be expressed by the original text according to a questioning and answering mode;

and a translation generation module 14 for converting the source language constituting the original text into the target language according to the determined single meaning.

In some embodiments of the present invention, the recognition module 10 is specifically configured to recognize that the original text has multiple meanings when recognizing a word having multiple meaning and/or multiple parts of speech in the original text, and/or when recognizing that the original text has a grammatical structure error or ambiguity.

In some embodiments of the present invention, the questioning module 12 is specifically configured to use the source language to ask a question; receiving an answer described in a source language according to a proposed question; determining a single meaning to be expressed by the original text based on the received answer.

Optionally, the presenting a question using the source language includes: analyzing the cause of the multiple meanings; determining the problem according to the reason; the problem is posed using the source language.

Optionally, the receiving an answer described in the source language according to the proposed question includes one or more of:

receiving a selected single meaning from a plurality of meanings described in the source language;

confirming the single meaning of the original text by adopting the source language prompt;

and modifying the original text by adopting the source language prompt.

In some implementations of the embodiment of the present invention, the questioning module 12 is further configured to, after receiving an answer described in the source language according to a question to be asked, take the received answer as an answer corresponding to the question to be asked, and store the answer in the answer database; determining the answer preferred by the user according to the received answer, taking the preferred answer as the answer corresponding to the question and storing the answer in the user preference database; and carrying out statistical analysis and/or machine learning on the received answers, obtaining machine answers through the statistical analysis and/or the machine learning as answers corresponding to the questions, and storing the machine answers into a statistical analysis and machine learning database.

In some implementations of embodiments of the present invention, the questioning module 12 is further configured to search the answer database, the user preference database, and the statistical analysis and machine learning database for an answer corresponding to a question to be asked before converting the source language constituting the original text into the target language according to the determined univocal meaning;

extracting answers when the answers are found, and determining the single meaning to be expressed by the original text according to the extracted answers;

and when the original text is not found, determining the single meaning to be expressed by the original text according to a question-answering mode.

The embodiment of the invention is a device embodiment corresponding to the first embodiment, and specific implementation can be referred to the first embodiment, so that the invention has corresponding technical effects.

A third embodiment of the present invention provides a translation apparatus including: a memory and a processor, the memory storing a translation program, the translation program when executed by the processor implementing the steps of the method according to any one of the embodiments. The translation device in the embodiment of the invention can be a computer, a mobile terminal, a wearable device, a tablet computer, a translation pen and the like.

A fourth embodiment of the present invention provides a computer-readable storage medium having a translation program stored thereon, the translation program being executable by at least one processor to implement the steps of the method according to any one of the first to the second embodiments.