How to parse requirements using a trained statistical model

From WeizmannWiki
Jump to: navigation, search

You may have chosen to create a grammar file based on previously created LSC examples (as described in How to train a model for statistical play-in). In order to use this grammar for parsing your requirements document, you should upload the file and tune several meta-parameters. To do this follow these steps:

  1. Upload the newly created grammar file:
    1. Select the LSC project and choose Properties -> PlayGo --> Play-In.
    2. Select the 'Statistical NL' tab and choose the newly created statistical grammar file:

    3. Choose whether or not you would like the sentence context to be taken into account (default: with context).
    4. Choose the number of candidates for each sentence you would like the statistical model to consider (default: 16 candidates).

Now, when you upload a requirements document, the newly created grammar will be used.