This is a useful moderation tool, but it has poor support for non-English characters, and the presence of non-English characters can even affect the detection of English and numbers. Here, taking the Simplified Chinese word “测试” (Test) as an example, the Watched words list contains three elements: “测试”, “Test”, and “123”. In the test below, none of the three examples triggered Watched words.
I searched within the site and found another similar issue about Censored words: Censored words do not respect word boundaries in non-latin alphabet. It seems that this is a common problem across the entire watch word matching system?