unilm icon indicating copy to clipboard operation
unilm copied to clipboard
verified publisher icon microsoft

About removing abstract noun phrases in GRIT construction

Open davidluciolu opened this issue 2 years ago • 4 comments

Hi! It is mentioned in the paper that

We eliminate certain abstract noun phrases that are challenging to recognize in the image, such as “time”, “love”, and “freedom”, to reduce potential noise.

So, the abstract noun phrases are eliminated manually or using spacy? Many thanks!

### Tasks

davidluciolu avatar Jan 04 '24 08:01 davidluciolu

Hi, The abstract noun phrases were eliminated manually. More specifically, we utilized ChatGPT to generate a list of such abstract nouns as candidates, and then we manually removed them from our dataset.

pengzhiliang avatar Jan 04 '24 12:01 pengzhiliang

Hi! Could you share the ChatGPT prompt? I am trying to use ChatGPT to generate a list, but results are awful. ChatGPT keeps generating repeated words.

Many thanks!

davidluciolu avatar Feb 28 '24 07:02 davidluciolu

Hi, @davidluciolu. Here are the abstract nouns we finally used:

abstract_nouns = ["time", "life", "love", "freedom", "happiness", "wisdom", "peace", "justice", "hope", "courage", "faith", "understanding", "advantage", "pursuit"]

You may use them to prompt ChatGPT to get more.

pengzhiliang avatar Feb 28 '24 10:02 pengzhiliang

thank you!

davidluciolu avatar Feb 28 '24 11:02 davidluciolu