FNatural

Minimal F# NLP library for functional text processing.

Structure

src/
├─ Core/
│  ├─ Tokenizer.fs
│  ├─ Stemmer.fs
│  └─ StopWords.fs
├─ Models/
│  ├─ TfIdf.fs
│  └─ NaiveBayes.fs
├─ Similarity/
│  ├─ Jaccard.fs
│  └─ Cosine.fs
└─ Data/
   ├─ Loaders.fs
   └─ Preprocessing.fs

Installation

dotnet add package FNatural

dotnet test tests/FNatural.Tests

Usage

Text preprocessing pipeline:

"Sample text" 
|> Preprocessing.cleanText
|> Tokenizer.tokenizeWords
|> StopWords.removeStopWords StopWords.Language.English

Sentiment classification:

let model = NaiveBayes.fit trainingData 1.0
"Positive experience" |> preprocess |> NaiveBayes.predict model

Document similarity:

let vec1 = TfIdf.vectorize tfidfModel doc1
let vec2 = TfIdf.vectorize tfidfModel doc2
Cosine.similarity vec1 vec2