Open Aupajo opened 13 years ago
Interesting. This would probably require some basic syntactic, as opposed to purely lexical, analysis. For example:
“My mind was set: I'd set up my train set tomorrow.”
‘Set’ is used as a participial adjective, a partial phrasal verb, and a noun being modified adjectivally by another noun. This is admittedly pretty pathological, but it demonstrates the complexity of this feature...
English is impossible to codify. I'm not out to build a hugely complicated lexical parser, just do some simple analysis as a starting point for the user.
Let the user pick and choose which words are nouns or not, on a one-by-one basis. The first-pass analysis gives them a guide, they can then touch it up. They're going to be a much better judge than any system could.
A nice way to adjust an analysis's deconstruction.
The scenario is that you're on a web-page or maybe a command line. You paste your text, and the analysis runs and shows you what it thinks the nouns are. You can select nouns it missed or deselect words that it incorrectly identified as nouns, before running the substitution.
Haven't thought much about what this would look like. Really, it just needs to handle two cases:
Although in my web-page example that would most likely be handled by the client side, so the simplest thing for now would simply be to implement
Analysis.from_deconstruction(deconstruction, options = {})
.