![](\"https:\/\/i0.wp.com\/www.albertopasca.it\/whiletrue\/wp-content\/uploads\/2018\/08\/Schermata-2018-08-08-alle-20.06.45.png\")
<\/p>\n
Use the Natural Language framework to perform tasks like language and script identification, tokenization, lemmatization, parts-of-speech tagging, and named entity recognition. You can also use this framework with\u00a0Create ML\u00a0to train and deploy custom natural language models.<\/p><\/blockquote>\n<\/div>\n<\/div>\n<\/section>\n
<\/p>\n<\/div>\n<\/div>\n
This framework provides a\u00a0high-level API for lots of language detection features using text.<\/p>\n
Let’s see some example:<\/h6>\n
\u3053\u3093\u306b\u3061\u306f\u3001\u79c1\u306f\u30a2\u30eb\u30d9\u30eb\u30c8\u3067\u3059\u3001\u79c1\u306f\u30a4\u30bf\u30ea\u30a2\u306b\u4f4f\u3093\u3067\u3044\u3066\u3001\u79c1\u306f\u4e0d\u660e\u306a\u8a00\u8a9e\u3067\u66f8\u3044\u3066\u3044\u307e\u3059\u3002<\/pre>\n\n
- Which language is this?<\/li>\n
- How many words contain this phrase?<\/li>\n
- There are names inside? Places? Company names?<\/li>\n<\/ul>\n
Who knows. Me not.<\/p>\n
Answers:<\/p>\n
\n
- JA<\/strong>, or better JAPANESE<\/li>\n
- 23<\/strong> words in this phrase<\/li>\n
- The english\u00a0translations is “Hi, I’m Alberto<\/strong>, I live in Italy<\/strong> and I write in an unknown language.”, so yes, there is a name and a place inside.<\/li>\n<\/ol>\n
\nLet’s do it to the new iOS\/macOS common framework, NLNaturalLanguage<\/strong> to see how it works.<\/p>\n
Examine this phrase:<\/p>\n
let<\/span> string = <\/span>\"Ciao, sono Alberto, vivo in Italia e scrivo in<\/span> an unknown language<\/span>. Mi piace la Coca-Cola.<\/span>\"<\/pre>\nit’s mixed, ITALIAN<\/em> \/ ENGLISH<\/em>. We can use this as a good\u00a0example.<\/p>\n
DETECTING LANGUAGE(s)<\/h6>\n
import NaturalLanguage\u2029\u2029\r\n\r\nlet string = \"Ciao, sono Alberto, vivo a Bergamo e scrivo in an unknown language. Mi piace la CocaCola.\"\r\n\r\n\u2029\u2029\/\/ create a new recognizer\u2029 \r\nlet languageRecognizer = NLLanguageRecognizer()\u2029\u2029 \r\n\/\/ that should read your string\u2029 \r\nlanguageRecognizer.processString(string)\u2029\u2029 \r\n \r\n\/\/ get eventually any language hypoteses\u2029 \r\nlet hypoteses = languageRecognizer.languageHypotheses(withMaximum: 2) \/\/2\u2029\u2029 \r\n \r\n\/\/ get the dominant language of the phrase\u2029 \r\nlet language\u00a0 = languageRecognizer.dominantLanguage!.rawValue \r\n \r\n\u2029\u2029print(\"First language is\u00a0 : \\(language)\")\u2029 \r\nprint(\"Other languages are: \\(hypoteses)\")<\/pre>\noutput in console:<\/p>\n
First language is<\/span>\u00a0 <\/span><\/span>: it<\/span>\u2029Other languages are: \r\n[__C.NLLanguage(_rawValue: <\/span>it<\/strong><\/span>): <\/span>0.9752411842346191<\/strong><\/span>, \r\n__C.NLLanguage(_rawValue: <\/span>en<\/strong><\/span>): <\/span>0.009950380772352219<\/strong><\/span>]<\/span><\/pre>\nWe receive the languages and the percentage of the confidence. Good. Italian is about 0.97%<\/strong> so we can trust\u00a0the algorithm.<\/p>\n
\nTOKENIZE A TEXT:<\/h6>\n
Let’s count the words (or the paragraph, or the sentences, or the document…) using\u00a0NLTokenizer<\/span><\/a><\/strong>:<\/p>\n
\/\/ create a new tokenizer<\/span>\u2029 \r\n\/\/ choose your unit (word, paragraph, sentences, document)<\/span>\u2029 \r\nlet<\/span> tokenizer = NLTokenizer<\/span>(unit: .word<\/span>) \r\n \r\n\u2029\u2029\/\/ set your language (or use the discovered one...)\u2029 \r\ntokenizer<\/span>.<\/span>setLanguage<\/span>( .<\/span>italian<\/span> ) <\/span>\/\/NLLanguage(language) )<\/em>\u2029\u2029 \r\n \r\n\/\/ link your string \r\n\u2029tokenizer.<\/span>string = <\/span>string\u2029\u2029 \r\n \r\n\/\/ get tokens\u2029 \r\nlet<\/span> tokens = <\/span>tokenizer.<\/span>tokens<\/span>(for: <\/span>string.<\/span>startIndex..<<\/span>string.<\/span>endIndex)<\/span>\u2029\u2029 \r\n \r\nprint<\/span>( <\/span>\"Words: <\/span>\\<\/span>(<\/span>tokens.<\/span>count)\"<\/span> ) \r\n<\/span>\u2029\/\/ <\/span>Words: 12 . <\/span><\/strong><\/pre>\nEXTRACT pieces of information:<\/h6>\n
Another cool feature is related to TAG<\/strong>, to extract tagged informations like, people names, city, places and organization names, using NLTagger<\/a><\/strong>.<\/p>\n
Let’s see how:<\/p>\n
\/\/ create a tagger\r\n<\/span>\u2029let<\/span> tagger = NLTagger<\/span><\/strong>(tagSchemes: [.nameType<\/span>])\u2029\u2029\r\n\r\n\/\/ set the text\r\n\u2029tagger.<\/span>string = <\/span>string\r\n\r\n\u2029\u2029\/\/ select the options\r\n\u2029let<\/span> options: <\/span>NLTagger.<\/span>Options = [<\/span>\u2029\u00a0 \r\n<\/span><\/span> .<\/span>omitPunctuation,<\/span>\u2029\u00a0 \r\n<\/span><\/span> .<\/span>omitWhitespace,<\/span>\u2029 \u00a0\r\n<\/span><\/span> .<\/span>omitOther,<\/span>\u2029 \r\n .<\/span>joinNames\u2029]\u2029\u2029\r\n\r\n\/\/ and the tag to extract\r\n\u2029let<\/span> tags: [NLTag<\/span><\/strong>] = [\u2029\u00a0\r\n<\/span><\/span> .<\/span>personalName,<\/span>\u2029 \r\n .<\/span>placeName,<\/span>\u2029 \r\n .<\/span>organizationName\u2029 \/\/ and much more...\u2029]\r\n\r\n\u2029\u2029\/\/ create all the tags\u2029\r\nlet<\/span> tags = tagger<\/span>.tags<\/span>(\u00a0<\/span><\/span>\u2029 \r\n in: <\/span>string.<\/span>startIndex..<<\/span>string.<\/span>endIndex,<\/span>\u2029 \r\n unit: .word<\/span>,\u2029 \r\n scheme: .nameType<\/span>,\u2029 \r\n options: options<\/span>) { tag, tokenRange in<\/span>\u2029 \r\n if<\/span> let<\/span> tag = tag, tags<\/span>.contains<\/span>(tag) {\u2029 \r\n print<\/span>(\"<\/span>\\(<\/span>tag.rawValue<\/span>) -> <\/span>\\(<\/span>string<\/span>[tokenRange])\"<\/span>)\u2029 \r\n }\u2029 \r\n return true\u2029\r\n}<\/pre>\nThe result is nice… with mixed languages happens something strange, but it’s ok.<\/p>\n
PersonalName -> <\/span>Alberto<\/strong><\/span>\u2029\r\nPlaceName -> <\/span>Bergamo<\/strong><\/span>\u2029\r\nOrganizationName -> an unknown language\r\n<\/span><\/strong>\u2029OrganizationName -> <\/span>CocaCola<\/strong><\/span><\/pre>\nEXTRA<\/h6>\n
You are able to know the language of the phrase so, you can easily speech the text in the real and in the correct language using AVSpeechSynthesizer<\/a><\/strong>!<\/p>\n
let speechUtterance = AVSpeechUtterance(string: string)\u2029\r\n\r\n\/\/speechUtterance.rate = 0.7\u2029\r\nspeechUtterance.volume = 1.0\u2029\u2029\r\n\r\n\/\/ set your discovered language\u2029\r\nspeechUtterance.voice = AVSpeechSynthesisVoice(language: language<\/strong>)<\/em>\u2029\u2029\r\n\r\nspeechSynthesizer.speakUtterance(speechUtterance)<\/pre>\nInstead of using these old techniques that make me laugh now… \ud83d\ude09<\/p>\n
[ObjectiveC] Text to Speech with Google Translate<\/a><\/p><\/blockquote>\n