NxNlp

Next-generation .NET NLP library for lexicon management, WordNet, spell checking, language detection, text segmentation, POS tagging, and more. Supports 66 languages with modular data packages.

Features

• Dictionary Lookup -- WordNet 3.0, Webster 1913, CC-CEDICT, Open Multilingual Wordnet (19 languages), AI-generated concise dictionaries
• Language Detection -- NTextCat (82 languages) + Catalyst neural detector fallback
• POS Tagging & Tokenization -- OpenNLP 2.5 Universal Dependencies models (36 languages) + ONNX transformer models
• Spell Checking -- Hunspell dictionaries for 56 languages
• Chinese NLP -- Jieba word segmentation, CC-CEDICT lookup, Chinese WordNet
• Japanese NLP -- MeCab morphological analysis, Japanese WordNet
• EPUB Parsing -- Extract text, metadata, TOC, and spine from EPUB e-books
• LLM Integration -- OpenAI-powered article generation, translation, word explanation
• Runtime Data Management -- On-demand download and installation of data packages from NuGet feed

Quick Start

Installation

# Core library
dotnet add package NxNlp --source https://nuget.easyrote.net/v3/index.json

# Language pack (includes all NLP data for a language)
dotnet add package NxNlp.LangPack.en --source https://nuget.easyrote.net/v3/index.json
dotnet add package NxNlp.LangPack.zh --source https://nuget.easyrote.net/v3/index.json

Or add the feed to your NuGet.Config:

<configuration>
  <packageSources>
    <add key="nuget.org" value="https://api.nuget.org/v3/index.json" />
    <add key="nxnlp" value="https://nuget.easyrote.net/v3/index.json" />
  </packageSources>
</configuration>

Language Detection

using NxNlp.Text;

var detector = new LangDetector(NxNlpDataPath.LangDetectDir);
var lang = await detector.DetectLangAsync("This is an English sentence.", CancellationToken.None);
// lang == "en"

var lang2 = await detector.DetectLangAsync("これは日本語の文です。", CancellationToken.None);
// lang2 == "ja"

Runtime Data Management

using NxNlp.DataManager;

var http = new HttpClient();
var feed = new NuGetFeedClient(http, "https://nuget.easyrote.net/v3/index.json");
var manager = new DataPackageManager(feed);

// List available packages
foreach (var pkg in manager.GetAvailablePackages())
    Console.WriteLine($"{pkg.PackageId}: {pkg.Description}");

// Install a data package at runtime
var progress = new Progress<DownloadProgress>(p =>
    Console.WriteLine($"{p.Phase}: {p.BytesDownloaded}/{p.TotalBytes}"));
await manager.InstallAsync("NxNlp.Data.Webster", progress);

// Query by feature
var required = manager.GetRequiredPackages("dictionary-wordnet");

Language Pack Discovery

using NxNlp.Common;

// List installed language packs
var languages = NxNlpLangPackManager.GetInstalledLanguages();
// e.g. ["en", "ja", "zh"]

// Check what's in a language pack
var contents = NxNlpLangPackManager.GetContents("zh");
// contents.HasUdModel, contents.HasHunspell, contents.HasOpenNlp25,
// contents.HasOmw, contents.HasJieba, contents.CcDictIds

// Get specific data paths
var udModel = NxNlpLangPackManager.GetUdModelPath("en");
var hunspell = NxNlpLangPackManager.GetHunspellPath("en");
var ccDict = NxNlpLangPackManager.GetCcDictPath("en", "en");

Data Packages

NxNlp uses a dual-track packaging system:

Module Packages (feature-scoped)

Install individual features as needed:

Language Packs (language-scoped)

One package per language, bundles everything needed for that language:

dotnet add package NxNlp.LangPack.en   # English
dotnet add package NxNlp.LangPack.zh   # Chinese
dotnet add package NxNlp.LangPack.ja   # Japanese
dotnet add package NxNlp.LangPack.de   # German
# ... 66 languages total

Each language pack may include:

Data Path Resolution

NxNlp looks for data files in this order:

1. NXNLP_DATA_DIR environment variable
2. NxNlp.Data/ under current working directory (NuGet contentFiles output)
3. Walk up directory tree to find AppData/_NxNlpData/
4. Legacy EasyOneX6 layout (backward compatibility)

To override, set the environment variable:

export NXNLP_DATA_DIR=/path/to/nxnlp/data

Requirements

• .NET 10.0 or later
• Git LFS (for cloning the repository -- OpenNlp25 and Onnx models)

NuGet Feed

All packages are hosted on:

https://nuget.easyrote.net/v3/index.json

License

Copyright EasyRote. All rights reserved.