ayanonagon / parsimmon Goto Github PK

View Code? Open in Web Editor NEW

707.0 23.0 46.0 220 KB

Parsimmon is a wee linguistics toolkit for iOS written in Swift.

License: MIT License

Ruby 1.73% Swift 98.27%

parsimmon's Introduction

Parsimmon

Parsimmon is a wee linguistics toolkit for iOS written in Swift.

We currently support Swift 2.0. If you are looking for Objective-C, please use version 0.3.4 or earlier.

Toolkit

Currently available tools:

Tokenizer
Tagger
Lemmatizer
Naive Bayes Classifier
Decision Tree (alpha)

Installation

The easiest way to get started is to use CocoaPods version 0.36 or higher. Just add the following line to your Podfile:

pod 'Parsimmon', '~> 0.5.0'

Examples

To start using Parsimmon:

import Parsimmon

Tokenizer

let tokenizer = Tokenizer()
let tokens = tokenizer.tokenize("The quick brown fox jumps over the lazy dog")
print(tokens)

(
The,
quick,
brown,
fox,
jumps,
over,
the,
lazy,
dog
)

Tagger

let tagger = Tagger()
let taggedTokens = tagger.tagWordsInText("The quick brown fox jumps over the lazy dog")
print(taggedTokens)

(
"('The', Determiner)",
"('quick', Adjective)",
"('brown', Adjective)",
"('fox', Noun)",
"('jumps', Noun)",
"('over', Preposition)",
"('the', Determiner)",
"('lazy', Adjective)",
"('dog', Noun)"
)

Lemmatizer

let lemmatizer = Lemmatizer()
let lemmatizedTokens = lemmatizer.lemmatizeWordsInText("Diane, I'm holding in my hand a small box of chocolate bunnies.")
print(lemmatizedTokens)

(
diane,
i,
hold,
in,
my,
hand,
a,
small,
box,
of,
chocolate,
bunny
)

Naive Bayes Classifier

let classifier = NaiveBayesClassifier()

// Train the classifier with some ham examples.
classifier.trainWithText("nom nom ham", category: "ham")
classifier.trainWithText("make sure to get the ham", category: "ham")
classifier.trainWithText("please put the eggs in the fridge", category: "ham")

// Train the classifier with some spam examples.
classifier.trainWithText("spammy spam spam", category: "spam")
classifier.trainWithText("what does the fox say?", category: "spam")
classifier.trainWithText("and fish go blub", category: "spam")

// Classify some new text. Is it ham or spam?
// In practice, you'd want to train with more examples first.
let firstExample = "use the eggs in the fridge."
let secondExample = "what does the fish say?"

print("\(firstExample) => \(classifier.classify(firstExample))")
print("\(secondExample) => \(classifier.classify(secondExample))")

'use the eggs in the fridge.' => ham
'what does the fish say?' => spam

License

MIT

Contributing

We’d love to see your ideas for improving this library! The best way to contribute is by submitting a pull request. We’ll do our best to respond to your patch as soon as possible. You can also submit a new GitHub issue if you find bugs or have questions.

Please make sure to follow our general coding style and add test coverage for new features!

parsimmon's People

Contributors

Stargazers

Watchers

parsimmon's Issues

Return confidence/probability percentage from classifyTokens

Hi,

Would be awesome if we could access a confidence level/percentage from ParsimmonNaiveBayesClassifer.classifyTokens method, maybe some easily-digestible human-friendly value.

Unify around a single legal boilerplate

I think people's Xcodes are set up differently...

Update docs for Swift

README
Code docs

update some image or gif preview?

Convert decision tree/node to use Swift

Other than the AppDelegate, that's the only remaining Objective-C code in the swiftify branch once #14 is merged.

Parsimmon doesn't work

i'm using Parsimmon to develop my application, but Parsimmon doesn't work when i build my app in device - iphone 5s, iOS 9.3.1

Swift 3.0

Parsimmon doesn't work with Swift 3.0

Classifier is not working as expected

Even though I write the exat same sentence, classifier can't identify string's category.

let prepared = ["file is packed",
            "Data Prepared",
            "information Arranged",
            "Record Created",
            "Data received"]
        let inhouse = ["Data is waiting",
            "Received at some location",
            "Received by somebody",
            "Package arrived"]
        trainCategory(prepared, category: "prepared")
        trainCategory(inhouse, category: "inhouse")

 let category = classifier.classify("Data is waiting")
        println(category) // it prints prepared

  private func trainCategory(items:[String], category:String) {
        for item in items {
            classifier.trainWithText(item, category:category)
        }
    }

PS: In Swift, documentation uses different syntax. For example following comment

    /**
        Returns the tokens for the input text using the specified linguistic tagger options.
        @param text Text to tokenize
        @param options Linguistic tagger options
        @return The tokens
    */

should be converted to

    /**
        Returns the tokens for the input text using the specified linguistic tagger options.

        :param: text Text to tokenize
        :param: options Linguistic tagger options
        :returns: The tokens
    */

For more information you may check http://nshipster.com/swift-documentation/

It's not working

Saving the state of Naive Bayes Classifier

Let's say I have huge amount of sentences to train Naive Bayes Classifier. Instead of doing this training every time in the program, is it possible to do it once save and reuse/add more training data later on?