Hello reader. I would like to briefly divert your attention from the news and stories with this technical article. So don't let this "clickbait" seed headline fool you.
In this article I will tell you how to make a parallel book, having two texts in different languages on hand. I wrote a web application that simplifies the alignment process by converting raw texts into books and parallel corpora. I want to share this project with the community, as well as get your opinion. I described the technical details here and here , the application code is open. Go.
Installation
I designed the application in the form of a docker container, so it shouldn't be difficult to run it on my machine. You can also run the application from source, the instructions are in the repository.
So, to run the application on your computer, you need to run the following commands:
docker pull lingtrain/aligner:habr docker run -p 80:80 lingtrain/aligner:habr
The container will download from the Docker Hub repository and run on port 80. Let's open the application in your favorite browser at localhost.
: , , .
:
, .
. 16 . , . - (, ) .
, .
, . , . , , , . . — ; . html .
:
| %%%%%author. | %%%%%author. | |
| %%%%%title. | %%%%%title. | |
| %%%%%qtext. | — , %%%%%qtext. | |
| %%%%%qname. | %%%%%qname. | |
| %%%%%image. | https://ya.ru/image.png%%%%%image. | |
| %%%%%h1. %%%%%h2. | ( ) | 1%%%%%h2. |
| %%%%%divider. | () | %%%%%divider. |
, , . , :
%%%%%author. %%%%%title. %%%%%h2. , , , , , , … , , .%%%%%qtext. . . %%%%%qname. . %%%%%h1. https://habrastorage.org/webt/2m/5r/jd/2m5rjdep6ej2rna5rhfioluig1u.png%%%%%image. %%%%%h2. : , : « ».%%%%%qtext. : ?%%%%%qtext. : … .%%%%%qtext. %%%%%qname. . , , , , . , . , . , . , , . . , . , , , , , . , . , , , : – ? ...
. , , .
, . .
, .
, . 10%, , — . , .
, , , — , .. , "". , .
. — , .
. 200 , . 5 , 1000 .
. "" , .
. - , . , .
, .
. , , . .
.
UI. ( ), — , . , . , . , , .
:
.
, .
, , .
, . — . , , Chrome .
, .
( ), ( ).
html . , — , . , .
. , , , , . , .
, TMX. - .
.
As a fan of language learning, I really want any person who studies or teaches languages to be able to quickly and easily make a parallel book for themselves. After all, finding one to your liking is more difficult than two books separately. Parallel enclosures are also of great benefit. For example, neural translation models are trained on them.
The project code is open source, the application is written in Vue and Python. Ideas and suggestions are welcome. Thanks for attention.