HALogen Demo

HALogen is a broad-coverage general-purpose natural language sentence generation system that combines symbolic rules with linguistic information gathered statistically from large online text corpora. It can be used to help do translation, summarization, and human-computer dialogue. See the ISI Natural Language Group home page.

Here is ONLINE DOCUMENTATION of the input representation.

The system software can be DOWNLOADED from here.

Try HALogen

Choose a meaning representation from this list: 

  (H9 / |have got,hold|
        :agent (h8 / HE)
        :patient (T9 / |talent,gift|
                     :mod (U9 / |extraordinary,unusual|)
                     :spatial-locating (E9 / |English language|)))

  (A1 / |possible>workable|
        :domain (A2 / |sell<cozen|
                    :agent i
                    :patient (T8 / |reliance,trust|
                                 :gpi they))
        :polarity negative)

  (h9 / |possible<latent|
        :domain (h2 / |obligatory<necessary|
                    :domain (e9 / |eat,take in|
                                :agent (y9 / YOU)
                                :patient (c9 / |poulet|))))

   A  minimally specified input at a syntactic level of abstraction

   An almost fully specified input at a syntactic level of abstraction

   Type in your own:

Which ngram model? Unigram Bigram Trigram
Track the best sentence for each possible sentence length? Yes No

*NOTE*: The trigram model and use of the length feature exponentially increase the amount of processing time. When using them, be prepared to wait 10 times longer or more.