Minimum description length for concepts should be a thing.
MP3 is a psychoacoustic algorithm for audio compression.
Could we develop a psycholinguistic method for textual compression?
Could we develop a psycholinguistic method for textual compression?
Ebook highlights, popular tweets, famous sayings, and the like provide datasets of sentences that have a particular ring to them.
Maybe you could parse them with the Stanford Parser or equivalent, compare to normal sentences, and find some differences?
Maybe you could parse them with the Stanford Parser or equivalent, compare to normal sentences, and find some differences?
There’s probably something in the broad area of text summarization that speaks to this, thought it seems a little different than just extraction- or abstraction-based summarization.
