In this article, I will introduce the second part of how to use the DeepL Translator API, which covers file translation. (Click here for Part 1) *Basic knowledge of Python and Linux is assumed. Translating text files Now that the warm-up is over, let's try file translation. This requires combining three API calls. Please prepare a text file named test.txt containing some Japanese text…

In this article, I will introduce how to use the DeepL Translator API in two parts: text translation and file translation. *Basic knowledge of Python and Linux is assumed. Please also refer to the official manual. https://www.deepl.com/en/docs-api Python 3.9 is used. Please note that the API specifications are as of July 2022. AP…

Kodensha Development Office Blog: Interview with the person in charge regarding the Japanese-English evaluation of "kode-AI Translation Cloud API"! This article is the second part, following the previous one. ～Click here for the first part of the blog～ Table of Contents (Part 2) 3. Evaluation Results: What is a BLEU score? 4. Discussion 3. Evaluation Results: What is a BLEU score? Satake: Before and after machine learning on 10,000 pieces of data, we used another 1,000 pieces of data that were not directly used for machine learning...

Table of Contents (Part 1) 1. What is machine learning for neural engines? 2. Introducing the evaluation procedure for machine learning results. Hello everyone! It has been a while (;^ω^) for the Kodensha Development Office blog. In this blog, our sales representative, Satake, conducts hands-on interviews with our developers to introduce various development and product information in a dialogue format.   The person who agreed to be interviewed today is this individual! This is Shibata-san, who is currently studying Chinese, Korean, and Russian. (Part 1…

Performance verification of domain-specific NMT engines. Using the IT field Chinese-Japanese parallel data collected in "I tried building a domain-specific neural machine translation engine: Focusing on data collection and performance verification (Part 1)," I actually prototyped domain-specific Japanese-Chinese and Chinese-Japanese NMT engines. I used a tool called OpenNMT to build the translation engine model. Now, let's actually translate test sentences (IT field) and check the performance of the IT field NMT...