pattern_syntax.md

Pattern syntax

Pattern consists of ordinary text and tokens that will be replaced during generation

ordinary [token] text

Each token consists of two parts:

• wm - Word mode (optional)
• ps - Part of Speech

and looks like this: [wm ps], or this: [ps]

Part of Speech

ps can be one of four supported PoS'es:

• verb
• adverb
• adj or adjective
• noun

By default, token will be replaced with random word of selected PoS

ps can also be followed by parentheses that contains a query for word lookup, then token will be replaced with random lookup result

Token examples:

[verb]
[adj(query)]

Word mode

Word in terms of WordNet isn't just a single word, but possibly a group of words or a phrase. Word mode helps to handle situations, where you need exactly one word in place of token.

This is an example phrase, or a WordNet word that will be used to provide examples in mode list below:

golden fairy lantern

So, the actual word modes (wm) list:

• First Word fw: golden
• Last Word lw: lantern
• Random Word rw: any of golden, fairy or lantern
• Separator sep(-): golden-fairy-lantern

fw and lw selects first or last word of a phrase accordingly, rw just selects random word from a phrase, and sep replaces space with separator provided in parentheses followed by it

Token examples:

[rw noun]
[lw adverb]
[fw noun(grass)]
[sep(_) noun]
[sep(-) verb]
[sep(::) adj(ironic)]