Using Boxer (further examples)

Producing DRSs for more than one sentence

It is possible to produce DRS for more than one sentence at a time, and indeed, it is very simple. What you do is create a file with the sentences you want to parse. Start every new sentence on a new line. The C&C parser will output CCG derivations for each sentence and when given this to Boxer, separate DRSs will be generated for each sentence. OK -- but what if you want to produce a DRS spanning several sentence? Just read on.

Producing a DRS spanning several sentences

By default, Boxer assigns a DRS for each sentence provided in the input file. Say you want to parse a text of two sentences, but would like Boxer to construct a DRS spanning both sentences, the way to do this is by using the <META> tags in the input file of the CCG-parser. The tags need to be followed by an identifier (not necessarily unique), and can therefore be used as a hook to introduce identifiers in DRSs, which is sometimes useful. Maybe you want to call the first text "Text 1" and the second "Text 2", as in the following example:

<META>'Text 1'
Vincent met a woman .
She was dancing .
<META>'Text 2'
Mia met a man .
He was dancing .

These identifiers will be output in the DRSs produced by Boxer. Note that identifiers need to be Prolog atoms. To be on the safe side, always add single quotes around identifiers as in the example input above.