Does there exist a dataset which contains a list of subject-verb-object triplets or adverb-verb/adjective-noun pairs that occur within a certain (large) corpus? If not, what would be the most efficient way to extract this information given a large corpus in text format?
You could use a parsing library like spaCy to parse sentences and then just look in the resulting syntax tree for your desired subtrees.
This assumes the corpus is mostly in English or another language supported by spaCy and reasonably clean.
Another approach would be to build n-grams and then pass the n-grams to a parsing library, but there are some problems with that.