So, yes, GPT should be able to add data to an excel file.
The easiest method i can think of off the top of my head would be to include some empty columns, namely a “type” and a “tag”.
Context is key, and specificity, clarity, and details will always help the model provide a better response, so definign what your keyword types mean would be very handy. You should also specify (just in case) that GPT picks one type for the column.
request it perform this logic line-by-line, going through each keyword and assessing it that way, you should be good.
The best thing to keep in mind for prompting this is to express the logic flow you want in as much detail as possible.
Let us know how it goes and if you need more guided help!
Yes you should. You can use a list of example words (in English) of how you would those like to be categorized. Its not like those three words have some universal meaning that makes it 100% clear where any given word should go. You might even have a similar data set in English (or another language) that already went through the classification. Paste relevant parts into the instruction.
I would NOT translate, first of all because for ChatGPT it is no problem to work with the words (and Brand names should not be translated, which is likely to happen if you did). And to @Macha 's point - I would consider asking it to make three columns per word -with (info/comm/brand) and let it put (No/Possible/Likely) or whatever you can play with.
I would ALSO just put this all in an Assistant and then just feed keyword in the prompt. You can easily play with it that way , even in the sandbox. All to get your prompt right. Do NOT try to let it work on all 600 keywords right away.
Let it play with 20 or 30. THe assistant instructions should have A LOT OF detail and examples AND the instruction to create the results in TABLE etc.
In the prompt you would just say 'process these keywords: …