CN103729346B - Method for dynamically generating mass language assets in multiple language industry standard formats - Google Patents

Abstract

The present invention is a method for dynamically generating massive language assets in a multilingual industry standard format, including: reading out the content in corpora and term bases in XML-based standard formats such as TMX and TBX by developing a parser and importing them into a specified In the database; while importing, it will automatically match and place database tables with the same content and different language pairs, and automatically generate a multilingual database with one source text and multiple sentences matching the target language; when the user uses it, according to the language specified by the user Yes, the searched results are automatically fed back to the user in the form of translation memory, and presented to the end user in a specific format for reuse; when adding or updating a multilingual database, it will automatically update relevant content in multiple languages, thereby ensuring language assets After the buzz, you can continue to let users get the updated translation memory content. Directly reuse the language assets saved in the text database format, the data is not easy to be damaged and lost, and the asset security is improved.

Description

Method for dynamically generating massive language assets in multilingual industry standard format

technical field

The invention relates to a method for dynamically generating massive language assets in a multilingual industry standard format, which is used for the development and application of a TM module in CAT software or a multilingual translation system, and belongs to the technical field of multilingual machine translation.

Background technique

TM (Translation Memory) is one of the technologies widely used in the field of computer-aided translation (CAT). With the help of TM technology, translation efficiency can be significantly improved and content consistency can be ensured. Due to the wide variety of CAT software developed using TM technology, the storage format of TM content varies greatly. In order to facilitate the exchange of TM data between translation agencies and CAT tools, an open standard called TMX (Translation Memory eXchange) has been successfully applied locally. culture and translation industry.

In the process of software and website localization translation, the content of the data files to be processed is relatively repetitive. In addition, because the content is updated frequently and is based on the update of the previous version, only a small amount of new content has been added or the original content has been modified. Minor corrections have been made, so it is necessary to take full advantage of what has already been translated in previous versions without having to re-translate.

TM technology effectively reuses the translated content. It uses segments (Segment) and TM library to improve translation efficiency. The translation database uses "Translation Unit" as the data unit, and combines each sentence in the source language with Sentences in the target language establish a corresponding link relationship. When translators use TM's CAT tool to translate content, the CAT tool continuously stores the latest translated content in the TM library. For the content to be translated (such as words, phrases, sentences, paragraphs), it first searches the TM library for whether the content is There is matching content, and the closest translation is automatically provided, and translators can easily insert the best matching translation.

With the continuous enrichment of translation content, the capacity of TM library continues to increase, translators do not have to worry about re-translating the same content again, they only need to focus on new content that needs to be translated, and the accuracy of TM can also guarantee the same content Translation consistency. This is the goal pursued by adopting TM technology.

However, with the continuous deepening of economic globalization, the localization and globalization of software/website industries are developing rapidly. Correspondingly, there are more and more localization tools and TM tools developed using TM technology, but these tools are Developed by different manufacturers, each has its own file data storage format. In addition, for a localization service agency, it often provides localization translation services for different customers or different projects of the same customer. Since different customers and different projects need to use different localization tools, it is often due to the lack of file data of each localization tool. A standard format can be exchanged, therefore, it is difficult to reuse previously accumulated TM library resources. Obviously, the standard format of TM library needs to be unified urgently.

To sum up, with the continuous deepening of economic globalization, the localization and globalization of software/website industries are developing rapidly. In addition to reusing the existing stored language assets (TM and terminology resources) in TMX and TBX formats, it is helpful To improve output and quality, reduce costs. Usually TMX or TBX appears in the form of a language pair, such as English to Chinese, English to German, etc. However, the technology in the industry still stays in the situation where a single language pair format is supported, and there is no technology to automatically generate a multilingual language pair from the same content in an existing single language pair.

Disadvantages of the existing technology: 1) The existing language asset storage architecture is two-dimensional and one-way, and the corresponding relationship between the source language and each target language cannot be opened; Automatic acquisition of multi-language (multi-dimensional) and multi-directional language pairs for content results in a great waste of resources, and if it needs to be acquired, it will inevitably result in huge labor costs.

Contents of the invention

In order to solve the above problems, the present invention aims to provide a method for dynamically generating massive language assets in a multilingual industry standard format. Technical scheme of the present invention is as follows:

A method for dynamically generating massive language assets in a multilingual industry standard format, comprising the following steps:

1. Read and import the content in the corpus and term base of TMX, TBX and other XML-based standard formats into the specified database through the development of the parser;

2. While importing, it will automatically match and place database tables with the same content and different language pairs, and automatically generate a multilingual database with one source text and multiple sentences matching the target language;

3. When the user is using it, according to the language pair specified by the user, the search results are automatically fed back to the user in the form of translation memory, and presented to the end user in a specific format for reuse;

4. When adding or updating a multilingual database, the relevant content in multiple languages will be automatically updated to ensure that after the language assets are dynamically updated, users can continue to obtain updated translation memory content.

The method for dynamically generating massive language assets in a multilingual industry standard format as described above, as a preferred solution: also includes:

Adopt λ corpus analysis module to provide analysis of industry standard formats TMX and TBX, read corpus information (including source language, target language, etc.) into memory, and convert them into binary objects;

Adopt the lambda corpus adaptation module to provide the matching function for the intermediate language corpus, and store the corresponding target language corpus in the correct position of the multilingual corpus matrix;

The λ corpus generation module is used to read the corpus information in the multilingual corpus matrix, and output it as a TMX or TBX format file according to industry standards, which is convenient for archiving and backup corpus or for other tools compatible with TMX or TBX.

The method for dynamically generating massive language assets in the multilingual industry standard format of the present invention has the beneficial effect that the language assets existing in the form of a multilingual database are physically independent of the language assets existing in the TMX and TBX formats, even if multilingual The deletion of the database will not affect the original language assets, thereby ensuring the security of the assets; moreover, the assets are stored on the storage medium in the form of text XML (TMX and TBX are both based on XML), which is different from the CAT tool The stored binary database files are frequently read, and their security can be guaranteed without accidental loss.

Direct processing of two industry standard formats, TMX and TBX, can bring the following beneficial effects:

1) Directly reuse the language assets saved in the text database format, the data is not easy to be damaged and lost, and the asset security is improved.

2) No need to manually convert the format, and automatically import the industry standard format to realize the reuse of language assets.

3) Automatically obtain multilingual and multidimensional language pairs and term pairs. For example, there are 3 language pairs in the original corpus. Through the invention of should use, additional value-added of assets can be realized, and 9 additional language pairs of corpus can be obtained, so as to make full use of language assets. For the globalization and internationalization of the enterprise's products, it maintains the consistency of language expression in the process of globalization, directly brings about the improvement of efficiency and quality, saves huge multilingual production costs, and shortens the time spent on the globalization of enterprise products. Time period.

4) Support high-speed query/reuse of massive multilingual assets.

Description of drawings

Figure 1. System block diagram of a method for dynamically generating massive language assets in a multilingual industry standard format.

specific implementation plan

Acronyms and key term definitions:

MTMM Multilingual Translation Memory Matrix multilingual translation memory matrix technology

TM Translation Memory translation memory

TU Translation Unit translation unit

TMX Translation Memory eXchange Translation Memory Exchange Format

TBX Term Base eXchange Term Base Exchange Format

CAT Computer Aided Translation Computer Aided Translation

LISA Localization Industry Standards Association Localization Industry Standards Association

OSCAR Open Standards for Container/Content Allowing Re-use Reusable Container/Content Open Standards

Specific examples are as follows:

The method for dynamically generating a large amount of language assets in a multilingual industry standard format includes the following steps:

1) By developing a parser, read out the content in the corpus and term base based on XML-based standard formats such as TMX and TBX and import it into the specified database;

2) While importing, it will automatically match and place database tables with the same content and different language pairs, and automatically generate a multilingual database with one source text and multiple sentences matching the target language;

3) When the user is using it, according to the language pair specified by the user, the searched results are automatically fed back to the user in the form of translation memory, and presented to the end user in a specific format for reuse;

4) When adding and updating a multilingual database, the related content in multiple languages will be automatically updated, so as to ensure that after the language assets are dynamically updated, users can continue to obtain updated translation memory content.

A method for dynamically generating massive language assets in a multilingual industry standard format, specifically including:

Adopt λ corpus analysis module to provide analysis of industry standard formats TMX and TBX, read corpus information (including source language, target language, etc.) into memory, and convert them into binary objects;

Adopt the lambda corpus adaptation module to provide the matching function for the intermediate language corpus, and store the corresponding target language corpus in the correct position of the multilingual corpus matrix;

The λ corpus generation module is used to read the corpus information in the multilingual corpus matrix, and output it as a TMX or TBX format file according to industry standards, which is convenient for archiving and backup corpus or for other tools compatible with TMX or TBX.

Language assets in the form of multilingual databases are physically independent of language assets in TMX and TBX formats. Even if the multilingual databases are deleted, the original language assets will not be affected, thereby ensuring the security of assets; Moreover, the assets are saved on the storage medium in the form of text XML (TMX and TBX are both based on XML), which is different from the binary database files that are frequently read and stored by CAT tools, and its security can be guaranteed without accidental loss.

Conceptual example sentences of the present invention:

A. Examples of the concept of translation memory (TMX):

Examples of single-language paired 2D TM content under normal circumstances:

English en-us: People’s Republic of China is a permanent member of the United Nations Organization

Chinese zh-cn: The People's Republic of China is a permanent member of the United Nations Organization

English en-us: People’s Republic of China is a permanent member of the United Nations Organization

French fr-fr: République populaire de Chine est member permanent de l’Organisation des Nations Unies

English en-us: People’s Republic of China is a permanent member of the United Nations Organization

German de-de: Der Volksrepublik China ist Mitglied der Organization der Vereinten Nationen

Through the technology of the present invention, any matching multilingual and multidimensional language pair TM will be automatically obtained, such as:

Chinese zh-cn: The People's Republic of China is a permanent member of the United Nations Organization

French fr-fr: République populaire de Chine est member permanent de l’Organisation des Nations Unies

Chinese zh-cn: The People's Republic of China is a permanent member of the United Nations Organization

German de-de: Der Volksrepublik China ist Mitglied der Organization der Vereinten Nationen

French fr-fr: République populaire de Chine est member permanent de l’Organisation des Nations Unies

German de-de: Der Volksrepubl ik China ist Mitglied der Organization der Vereinten Nationen

B. For example, explain the concept of term base (TBX):

The content of monolingual two-dimensional terms in common cases:

English en-us: Computer-assisted translation

Chinese zh-cn: Computer Aided Translation

English en-us: Computer-assisted translation

French fr-fr: Traduction assistée par ordinateur

English en-us: Computer-assisted translation

German de-de: Computerrunterstützte

Through the technology of the present invention, any matching multilingual and multidimensional language pair terms will be automatically obtained:

Chinese zh-cn: Computer Aided Translation

French fr-fr: Traduction assistée par ordinateur

Chinese zh-cn: Computer Aided Translation

German de-de: Computerrunterstützte

French fr-fr: Traduction assistée par ordinateur

German de-de: Computerrunterstützte

Direct processing of two industry standard formats, TMX and TBX, can bring the following beneficial effects:

1) Directly reuse the language assets saved in the text database format, the data is not easy to be damaged and lost, and the asset security is improved.

2) No need to manually convert the format, and automatically import the industry standard format to realize the reuse of language assets.

3) Automatically obtain multilingual and multidimensional language pairs and term pairs. For example, there are 3 language pairs in the original corpus. Through the invention of should use, additional value-added of assets can be realized, and 9 additional language pairs of corpus can be obtained, so as to make full use of language assets. For the globalization and internationalization of the enterprise's products, it maintains the consistency of language expression in the process of globalization, directly brings about the improvement of efficiency and quality, saves huge multilingual production costs, and shortens the time spent on the globalization of enterprise products. Time period.

4) Support high-speed query/reuse of massive multilingual assets.

Every manufacturer hopes that users will be more dependent on their own CAT products, but from the perspective of users, a method that supports massive language assets and automatically generates multi-language pairs from the same content of a single language pair ensures asset security. It will be quite valuable to realize the maximum application of resources. By adopting the technical solution of the present invention, beneficial results can be obtained: in addition to ensuring the reuse of the original single-language sentence pairs and asset security, at the same time, automatically obtaining multi-language and multi-dimensional language pairs for users, realizing additional value-added assets and maximizing language assets efficacy.

The above descriptions are only preferred embodiments of the present invention, and any non-creative improvements made by those skilled in the art around the spirit fall within the scope of protection of the present invention.

Claims (2)

1. The method for dynamically generating a large amount of language assets in a multilingual industry standard format is characterized in that: comprising the following steps: (1) reading content in corpus and term bases based on XML standard formats such as TMX and TBX by developing a parser Take it out and import it into the designated database; (2) while importing, it will automatically match and place database tables with the same content and different language pairs, and automatically generate a multilingual database with one source text and multiple sentences matching the target language; ( 3) When the user is using, according to the language pair specified by the user, the searched results are automatically fed back to the user in the form of translation memory, and presented to the end user in a specific format for reuse; (4) When adding and updating the multilingual database , the multilingual related content will be automatically updated to ensure that after the language assets are dynamically updated, users can continue to obtain updated translation memory content; The language assets are stored on the storage medium in the form of text XML, and the language assets A in the form of a multilingual database are physically independent from the language assets B in the formats of TMX and TBX. 2. the method for dynamically generating the massive language assets of multilingual industry standard format according to claim 1, is characterized in that: described step (1) specifically comprises: adopt λ corpus parsing module, provide industry standard format TMX and TBX Parse, read the corpus information into the memory, and convert it into a binary object; Described step (2) specifically comprises: adopt λ corpus adaptation module, provide the matching function to intermediate language corpus, and store corresponding target language corpus in the correct position of multilingual corpus matrix; Adopt λ corpus generation module, provide to read multiple The corpus information in the language corpus matrix is output as a TMX or TBX format file according to industry standards.