README.md

metadata

title: WiktionaryDE
emoji: 🐠
colorFrom: indigo
colorTo: pink
sdk: gradio
sdk_version: 5.49.1
app_file: app.py
pinned: false
license: cc-by-sa-3.0

🇩🇪 WiktionaryDE - German Linguistics Hub

An advanced multi-tool for German linguistic analysis that combines German Wiktionary database query with multiple morphological engines and semantic knowledge bases into a single, comprehensive interface.

🎯 Overview

This Space aggregates multiple German NLP tools and databases to provide:

• Deep morphological analysis of German words
• Contextual sentence analysis with semantic ranking
• Full inflection tables (declensions and conjugations)
• Thesaurus and semantic relation discovery
• Grammar and spelling checking

🛠️ Tools & Data Sources

Core Databases

• Wiktionary Database: 3.7GB cstr/de-wiktionary-sqlite-normalized database providing ground truth for lemmas, inflected forms, definitions, examples, and pronunciation
• OdeNet (WordNet): German thesaurus for synonyms, antonyms, hypernyms, etc.
• ConceptNet: Multilingual knowledge graph for semantic relations

Morphological Engines

• DWDSmor: High-precision FST-based analyzer from zentrum-lexikographie/dwdsmor-open
• HanTa: Hanover Tagger for robust morphological analysis and lemmatization
• spaCy-IWNLP: de_core_news_md combined with IWNLP for spaCy-based analysis
• Pattern.de: Full inflection table generation

Additional Tools

• LanguageTool: German grammar and spelling checks

📖 Main Features

1. Word Encyclopedia (DE)

The primary non-contextual tool for analyzing single words.

What it does:

• Finds all possible analyses (e.g., "Lauf" as noun vs. "lauf" as verb)
• Aggregates data from all engines and databases
• Cross-validates results to filter out artifacts
• Provides complete morphological, semantic, and inflectional information

Engine Options:

• Wiktionary (Default): Most accurate, database-driven
• DWDSmor: High-precision formal grammar
• HanTa: Robust tagger-based
• IWNLP: spaCy-based analysis

The engine selector automatically falls back to other engines if no result is found.

2. Comprehensive Analyzer (DE)

Full sentence analysis with contextual disambiguation.

Features:

• Uses spaCy to parse sentences and extract lemmas
• Runs full Word Encyclopedia analysis on each lemma
• Contextual Ranking: Uses sentence similarity to rank semantic senses by relevance to the full sentence
• Provides integrated analysis of all words in context

3. Individual Engine Tabs

Direct access to raw outputs from:

• Wiktionary
• DWDSmor
• HanTa
• IWNLP

Useful for comparing individual engine outputs.

4. Component Tools

Raw access to specialized tools:

• spaCy: Dependency parsing and NER
• Grammar: LanguageTool checking
• Inflections: Pattern.de inflection tables
• Thesaurus: OdeNet relations
• ConceptNet: Semantic knowledge graph

⚙️ Technical Details

• SDK: Gradio 4.31.0
• Database Size: 3.7GB (Wiktionary sqlite)
• Processing: Multi-engine pipeline with intelligent fallback
• (basic) Quality Control: Cross-validation between engines to filter artifacts

📝 License

The code for this Gradio interface is licensed under CC-BY-SA-3.0.

The underlying models and data sources retain their original licenses:

• Wiktionary: CC-BY-SA
• DWDSmor: Open license (zentrum-lexikographie)
• HanTa: Various open licenses
• spaCy models: MIT License
• OdeNet: CC-BY-SA
• ConceptNet: CC-BY-SA

Note: This is a simple educational tool and work-in-progress. Many results will not be consistent and faulty.