Written simulations make sense, but they're boring. Markov chains and similar text generators don't make a lot of sense, but they're interesting. So, I'm going to try to get the best of both worlds.
To that end, I'll try writing up a simple AI in a simulated environment, trying for life-or-death goals. Instead of describing what's being done directly, however, the parser will generate a relevant corpus for pt-voicebox, then attempt to output a descriptive sentence of what's occurring, using output from word2vec or WordNet to gauge how relevant a given word is to the current situation (moving down the list if a certain threshold is not met).
Goals if I have time: use of sentiment analysis for matching certain moods depending on the situation (fear, love, happiness, etc.); recognition of nouns, verbs etc. for less nonsensical sentence structure; tests to avoid repetitive sentences.
Written simulations make sense, but they're boring. Markov chains and similar text generators don't make a lot of sense, but they're interesting. So, I'm going to try to get the best of both worlds.
To that end, I'll try writing up a simple AI in a simulated environment, trying for life-or-death goals. Instead of describing what's being done directly, however, the parser will generate a relevant corpus for pt-voicebox, then attempt to output a descriptive sentence of what's occurring, using output from word2vec or WordNet to gauge how relevant a given word is to the current situation (moving down the list if a certain threshold is not met).
Goals if I have time: use of sentiment analysis for matching certain moods depending on the situation (fear, love, happiness, etc.); recognition of nouns, verbs etc. for less nonsensical sentence structure; tests to avoid repetitive sentences.
Well, we'll see how it goes!