I'm working on my chatbot program and it could really use an extension to its vocabulary. Currently it picks words out of a list of just a few hundred, all entries containing data like the word itself, the context this particular entry applies for, how polite, blunt, fancy or aggressive (and many other characterizing attributes on a scale of 0 to 1) it is. All painstakingly filled out by myself.

Since by my estimate, to cover the whole of commonly used English, the database would need at least 600,000 entries, doing that my myself would be madness. I'm looking for resources to pull on, anything that describes words with numbers or at least booleans (like for instance this word is nice and this one is not so nice). Doesn't matter if the format is ready to be fed into a number crunching machine, I can build my own parser for that. Thanks in advance.

  • It's not so much the words as the contexts they're in. Have you seen Wordnet and Framenet? Jun 23, 2014 at 19:50
  • I have and while they are impressive dictionaries, they are way too abstract to be useful to me. And yes, the context weighs heavily on the flavor of a word but I have to put a quantifiable limit to the complexity and the amount of the data and those 600k entries would include duplicates for various (albeit limited compared to humans) contexts.
    – user81993
    Jun 23, 2014 at 20:10
  • 4
    This question appears to be off-topic because it is a request for resources Jun 23, 2014 at 20:25


Browse other questions tagged or ask your own question.