Parse a pipeline configruation. This namespace supports a simple DSL for parsing a pipeline configuration (see zensols.nlparse.config). The configuration string represents is a component separated by commas as a set of forms. For example the forms:

zensols.nlparse.config/tokenize("en"),zensols.nlparse.config/sentence,part-of-speech("english.tagger"),zensols.nlparse.config/morphology

creates a pipeline that tokenizes, adds POS and lemmas when called with parse. Note the double quotes in the tokenize and part-of-speech mnemonics. The parse function does this by calling in order:

• (zensols.nlparse.config/tokenize “en”)
• (zensols.nlparse.config/sentence)
• (zensols.nlparse.config/part-of-speech “english.tagger”)
• (zensols.nlparse.config/morphology)

Soem configuration functions are parameterized by positions or maps. Positional functions are shown in the above example and a map configuration follows:

parse-tree({:use-shift-reduce? true :maxtime 1000})

which creates a shift reduce parser that times out after a second (per sentence).

Note that arguments are option (the parenthetical portion of the form) and so is the namespace, which defaults to zensols.nlparse.config. To use a separate namespace for custom plug and play To use a separate namespace for custom plug and play components (see zensols.nlparse.config/register-library) you can specify your own namespace with a /, for example:

example.namespace/myfunc(arg1,arg2)