1. Did you edit the ProjectTerms glossary in some external editor? Was it created in CafeTran? Is the language pair (see the green header after reloading) of the glossary the same as the project's language pair?
2. When you move to the next/previous segment, the glossary shows only the glossary terms present in that segment (highlighted in beige) or nothing if CafeTran can't find such terms.
3. The red dot indicates that the segment has been edited in the current session.
Probably, you haven't created a new project with the document for translation yet. You may be editing the translation memory (a .tmx file) only. You can attach the screenshot of the interface after loading the file to check it.
I hope it's just that and not anything more problematic. I tried many times to attach a screenshot of the interface to my post, but whenever I do, the post won't go through. I did create a new project at the beginning, and when I reopen the CafeTran app, the project shows up next to the Open button. There is also a Project file on my computer containing all the files. I completed the project through to the end (It was a small demo project) and exported it, so the Project file contains the exported bilingual file as well as the ProjectTerms.txt...
I've now found that sometimes a matched term appears in the ProjectTerms window (with the beige highlight in the source text) and sometimes it doesn't. There doesn't seem to be any pattern as to why a particular term sometimes shows up and sometimes doesn't. A term in the source text in segment 5, for example, will show up in the ProjectTerms window, but in the very next segment, this exact same terms will not show up.
Your glossary terms may appear in various forms in the sentence. Then please enable Prefix matching option in Edit > Preferences > Glossary tab to increase the chances of their matching.
I have figured out the issue. My source language is Korean, and Korean poses two problems: 1. it uses characters not letters; 2. the characters are joined together to create "words (a clump of characters, essentially)" which, in English (for example), would be the equivalent of joining multiple words together into a single word, e.g., "Thatis", "toschool", "onlyresults" and "doesn'tevenattempt". On the other hand, thousands of terms can consist of a single Korean character. "Cooked rice", "snow", "house", for example, all consist of a single Korean character. Therefore, in order to capture all the matching terms, I would need to select Prefix matching and set the Minimal prefix length to "1". However, this then creates the problem of capturing all the words consisting of two or more characters that start with one matching character. The first example that came up in my case was "하얼빈" and "하다". These two terms share the same first character, but in the former, this character is the first syllable of Harbin (a city in China) while in the latter, this character is the stem of the verb "to do". It's like matching "Harbin" with "happy" because they both start with "ha".
The best way to get around the issue would be to keep the Minimal prefix length to "1" and just ignore the bad matches, so it's not an insurmountable problem. I just thought it was worth flagging. Thank you for your help, Igor.