Open oraby8 opened 6 months ago
@oraby8 For non-English languages, finding a specific list of words or phrases is a challenging task. Internet corpora are not available for every language, and even when they exist, such as for Chinese
, they may not be large enough to identify reliable patterns.
While working with Large Language Models (LLMs) like ChatGPT
for non-English text generation, the focus shifts from detecting the presence of specific words to a different approach. This approach needs to be language-specific. For example, if you are targeting Arabic and trying to detect cheating in an essay about a famous personality, you should provide a separate document with extensive technical information about that person. Then, evaluate each submitted document and score them based on the amount of missing technical information.
The reason behind this method is that LLMs like ChatGPT tend to generate general information and often lack technical depth. Even if prompted to include technical details, LLMs may not provide the full range of knowledge that someone would provide through dedicated study and research.
I'm wondering how I could add a list for another language. Is there a particular method to do that ?