Reasons for inaccurate translation software

ArticleDirectory

• Phrase elimination Ambiguity
• First, there is a language followed by a syntax
• Improvement of translation software

Most people who have access to English may use Kingsoft, youdao, Geling, Google, Baidu, etc. there is generally no problem with translation words and phrases. however, the translation of a sentence or a whole article is not accurate enough in many places. in today's rapid development of the software industry, although language translation software has made some progress, it is also slow. accuracy is still not too high. so you have to ask why the translation is so inaccurate and so difficult?

There are three major reasons:

1. Multiple definitions of one word are prone to ambiguity

2. for cultural reasons, the same words do not mean the same in different cultures.

3. People often do not strictly follow the syntax to use the language.

 

Ambiguity of a word

At the beginning, the translation software usually adopts the syntax strictly. First, the entire dictionary is used. after a sentence is taken, the corresponding words in the sentence are replaced, and then the order of words is sorted according to the syntax rules. just like what the compiler is doing now, the compiler is actually a simple translation software.Source codeTranslation into a machine that can recognize other machine languages.ProgramThe language is concise and unambiguous. If there is any ambiguity, it will be treated as an error. natural Language is often ambiguous. For example, a word has many meanings, not just one meaning. A word in English usually has 10 or 20 meanings. the Chinese language is a little better, and the meaning of a word is not too much. In fact, the meaning of a word in ancient Chinese is quite a lot. If you have studied classical Chinese, you will be able to deeply understand it.

Phrase elimination Ambiguity

The specific meaning of a synonym depends on the context. Therefore, many commonly used phrases are usually formed.When several words are together, they are called phrases. With context, most of the ambiguities can be eliminated.. With the development of storage devices, you can store more phrases in translation software, not just the meaning of words. if so, the accuracy should be quite high. however, in fact, the accuracy of the translation methods is very low. this is especially true in everyday languages rather than academic articles. why?

 

Cultural Reasons

 

Many things cannot be understood under certain cultural backgrounds. the cultural factor is a relatively imaginary thing, and it is the inheritance of various customs and cultures. You cannot say you want to move a culture to a computer. for example, if there is a Chinese saying, I would like to greet your family of women. after English translation, the ladies who have always paid attention to gentleman manners give priority to the foreigners who think the Chinese are really polite, so respect for lesbians. even people cannot understand it very well. Do you expect the translation software to translate it for you?

In addition, dialect text cannot be translated into English even more like Chinese poetry. In the past, it was simply beyond sight. what do you mean after you translate it into English like Tang Shi and Song ci? It turns into a few nonsense words and loses its original charm.

 

Reasons for laziness

Another reason for inaccurate translation is that translation software is never accurate.When people use a language, they always do not fully follow the syntax. They always make syntax errors constantly, and we usually do not treat some sentences as incorrect sentences. We know thatProgramming LanguageIt is to first develop some syntax rules, and then the compiler does not all follow the syntax, but gives some of its own rules. however, in any case, there are definite rules for reference. then the programmer starts to write the sourceCodeIf you use the compiler to check whether there is a syntax error while writing it, there is no problem with the final code. The programming language has the syntax before it has the source code.

First, there is a language followed by a syntax

However, our natural language does not have syntax first. we have no strict syntax in China for thousands of years. even punctuation does not match, a maximum of a few points. the grammar and punctuation in our language courses were completed by many language masters in the Republic of China referring to the Grammar System in English.

So can you expect us to use the language in full compliance with the syntax? Although our education system was initially intended to trick us into completely following the syntax. what impressed us most when I was studying was the question of sentence judgment. however, as soon as the school came out, we threw aside those grammar rules. few people think about writing their own sentence when talking or writing articles. Of course, some professional editors, language scholars, authors, and legal rule makers are not here. therefore, we often unconsciously spread the sentence.

In addition, it is easy for people to get lazy. Some sentences can be omitted if they can be omitted, and some keywords are usually used. in addition, the numerous sentences in daily life do not bring us too much inconvenience. because people are very smart, you can know what you are talking about based on a few keywords.

Generally speaking, language is just a tool for us to communicate with each other. As long as we can achieve our goal, we do not have to do it completely by syntax.

 

Improvement of translation software

Since the accuracy of translation based on grammar is not very high, many smart people will naturally find a solution. therefore, an experience-based approach emerged. analyze a large number of existing language data. various books and articles. find out the matching rules of some statements and phrases. it mainly counts the probability that a word appears before or after another word. we have mentioned that a single word is ambiguous and can be used to eliminate ambiguity through common phrases. the concept of frequently-used phrases is expanded here. there are more frequently-used phrases (determine by probability if they are commonly used), and the previous combinations of each phrase are also taken into account.

Therefore, the current progress of translation software is mainly reflected in the collection and analysis of massive language data. in the future, you may save the meanings of some commonly used statements as a word. you don't need to perform conversion analysis. for example, if you swear in Cantonese, your lungs are saved as a word. for example, it means fuck you.