[We are @ciceron] We Introduce a machine translation system MARO, the first step to use translation data.
introduceyourself·@ciceron·
0.000 HBD[We are @ciceron] We Introduce a machine translation system MARO, the first step to use translation data.
<center></center>
Hi, everyone! @ciceron is back again with more interesting stuff to tell you!
Today we will show you as many as three pictures where we tried to summarize the [previous post](https://steemit.com/introduceyourself/@ciceron/we-are-ciceron-we-do-translation-for-connecting)! Quite special, isn’t it?
(~~※sorry for bad drawings※~~)
<center></center>
### About MARO
###### <br>
MARO is an **AI-powered machine translation system that is trained on data.** Everyone has heard about AI, haven’t you? MARO was trained on **1,589,772 words that passed through the hands of professional translators.** We do still turn to translators when we need something simple to be translated, however, even right now MARO is being trained to become a complete tool that can assist translators in future and help them increase their productivity and efficiency.
> AI-powered machine translation system? You say GoogXX is the best?
There are many machine translation systems including GoogXX, NavXX, MX and so on. However, **field-specific training** is what makes MARO developed by @ciceron different.
### Storing classified data
###### <br>
Up to now, when building a machine translation system, all words have been stored together in one dictionary. However, @ciceron trains MARO on **data classified and stored in 13 different categories** including medicine, natural sciences, law, liberal arts, literature, trade, politics, computer technologies and so one. Apart from the data classification conducted by @ciceron, Maro can also group the data into different categories. Since Maro stores words and data for each field and industry separately, will it be able to do more accurate and subject-oriented translations?
In fact, **the BLEU score which is used to evaluate the performance of machine translation systems** is indeed higher than those showed by other machine translation systems! ~~(this is true only for Korean - English language pair, though haha)~~
### Field-specialized translation system
###### <br>
Because it’s just an early version, it’s not a perfect field-specialized translation system, but it’s possible to combine classified data and create a model specifically for each separate field. **The ability to create a specialized model for a specific field or industry by combining necessary data is one of the advantages that MARO has.** For example, it is possible to generate a machine translation model that will work only with patents, fiction, art or business etc. by storing words that are used in the corresponding field.
<br>
---
Sorry for such a long introduction. We didn’t intend to boast so much haha
Next time we will show you what MARO looks like and explain how to use it :)
You are looking forward to it, aren’t you?
***to be continued...***
---
### Here are some tips how you can use services provided by @ciceron!
1. If there is a post you want to be translated, just tag @ciceron.
2. If there is a post you would like to introduce to foreign Steemians, also just tag @ciceron
3. If a post is really worth being translated, @ciceron might do it for you!