set optimal parameters for unsupported languages

[abubelinha]

Sorry I don't understand a word about how the Bayes / Odds options work in this context.

So I need to figure out what's the best way to go to detect scientific names in non supported languages, like French, Spanish or Portuguese. Would you change something in this example API query?

{
  "text": "Número.\nNom. Passer domesticus L.\nHabit. en las márgenes del Duero, con Quercus toza, Quercus suber L., etc.\nLegit P. Rousseau.\nPasseriformes",
  "noBayes": true,
  "oddsDetails": false,
  "language": "",
  "wordsAround": 0,
  "verification": true,
  "sources": [11,12,1]
}

BTW I suppose "sources" overrides "verification", right?
I mean, it only makes sense to use "verification" (true/false) when you don't provide a list of sources.

[dimus]

ATM language is more of a placeholder, and I would say it is save to keep defaults for any language. Only German and Eglish have Bayes training sets, and only English went through a thorough trial and error.