Shulou(Shulou.com)06/02 Report--

Machine translation (MT) is automatic translation, which is the process of translating text from one natural language to another using computer software.

Whether it is manual translation or machine translation, the meaning of the text in the source language must be fully restored in the target language, that is, translation. Although this may seem simple on the surface, it is actually much more complicated. Translation is not just word-for-word substitution, translators must explain and analyze all the elements in the text and understand the relationship between words. This requires extensive expertise in grammar (sentence structure) and semantics (meaning) in both the source and target languages, as well as familiarity with each language region.

Manual translation and machine translation have their own challenges. For example, any two independent translators cannot produce a consistent translation of the same text in the same language, and it may take several rounds of revision to satisfy the customer after the translation. Obviously, machine translation is more difficult to produce high-quality translation with high customer satisfaction.

Rule-based Machine Translation Technology

Rule-based machine translation relies on countless built-in language rules and millions of bilingual words for each pair of languages.

This technique parses the text and creates a transitional expression from which the text in the target language is generated. This process requires a wide range of word meanings, including morphological, syntactic and semantic information, as well as a large number of rules. This technique uses these complex grammar sets and then converts the syntax structure of the source language into the target language.

Rule-based MT ⬆ of "Curve Saving the Nation"

Translation is based on a large vocabulary and complex grammatical rules. Users can improve the quality of translation by adding terms in the translation process. Users can customize their vocabulary to override the default settings of the system.

In most cases, there are two steps: the company's initial investment significantly improves quality at a limited cost, and continuous investment to gradually improve quality. Although rule-based MT enables companies to meet the quality threshold and higher, the quality improvement process can be long and expensive.

Statistical machine translation technology

Statistical machine translation uses statistical translation model, and its parameters are derived from the analysis of monolingual and bilingual corpora. The construction of statistical translation model is a fast process, but this technique depends to a large extent on the existing multilingual corpus. A specific language needs at least 2 million words, and a general language needs more. In theory, it is possible to reach the quality threshold, but most companies do not have such a large number of existing multilingual corpora to build the necessary translation models. In addition, statistical machine transformations are CPU-intensive and require a wide range of hardware configurations to run average performance-level transformation models.

Comparison between Rule-based MT and Statistical MT

Rule-based MT provides good extraterritorial quality and is predictable in nature. Customizable vocabulary ensures improved quality and compliance with company terminology. However, the translation results may lack the fluency expected by readers. In terms of cost, the customization period required to reach the quality threshold can be long and costly.

Statistical MT can provide good quality when a large corpus is available. The translation is smooth and easy to read, so it meets the expectations of users. However, translation is neither predictable nor consistent. Excellent corpora are automatically generated and cheap. However, the training of the general language corpus, that is, the text outside the specified domain, has a worrying effect. In addition, statistical MT requires a lot of hardware to build and manage large translation models.

This article is reproduced from data Xinghe platform: https://www.bdgstore.com.cn/portal/article/index/id/167.html

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.