The first is the detection range of HowNet's master thesis:
There are only a few related English databases in the middle part, and there is no component of mutual translation detection here. Let's take a look at the test content:
Because of the content, I only copied part of a report, and that's about it. After reading my whole article, I didn't copy the reports of English literature. I also have about 20 reports from other friends, and there are no English documents.
My preliminary inference is that this technology is unlikely to be widely used: because there may be many sentences translated from English, it is difficult and inaccurate to match, which does not conform to the principle of (keyword) matching.