NPannotator

A flexible, semi-automatic noun phrase annotation tool for multilingual linguistic research and corpus annotation.

Overview

NPannotator is a browser-based application designed to streamline the annotation of noun phrases across any language. It features automated tag suggestions, customizable category hierarchies, and comprehensive export capabilities.

Key Strengths:

• 🌍 Language-agnostic — annotate noun phrases in any language
• 🏷️ Hierarchical tagging — add or remove categories, subcategories, and types
• 🧠 Smart suggestions — automatically suggests tags based on previously annotated phrases
• 💾 Flexible storage — lexicon and annotation history stored locally (no external API required)
• 📊 Multiple export formats — JSON, CSV, and Excel (XLSX)
• 🔄 Batch import — upload CSV or Excel files for annotation

Features

1. File Upload & Setup

• Import annotation data from CSV or Excel (XLSX) files
• Auto-detect column headers or manually assign them
• Define role mapping for data columns:
  • Data column (noun phrase text)
  • Context column (sentence or example)
  • Language (single language for the session)
  • Dataset code (corpus/source identifier)
  • Source (optional: document title, book reference, etc.)

2. Noun Phrase Annotation

• Tokenization — automatically splits phrases into tokens
• Multi-token selection — tag contiguous or non-contiguous sequences
• Interactive highlighting — see noun phrase locations within full context
• Flexible glossing — add word-level glosses
• Phrase-level translation — optional translation for the entire phrase

3. Tagging System

Hierarchical Tag Structure

By default, tags are organized in three levels:

• Category (e.g., NOUN, ADJ, ART, POSS, NUM, DEM, QUANT, RC, PP)
• Subcategory (e.g., NOUN → Animate/Inanimate)
• Type (e.g., NOUN-INANIM → Object/Event)

Auto-Suggestion

• Learns from previously tagged tokens and phrases
• Suggests tags and glosses when the same token sequence appears again
• Shows multiple suggestions if available
• One-click application of suggested tags

4. Lexicon Management

• Automatic lexicon building — lexical entries created as you annotate
• Meaning-based organization — multiple meanings per word with glosses
• Phrase tracking — each meaning records which phrases it appears in
• Edit & review — add/delete meanings, update glosses
• Language-aware — entries keyed by (language, word form) pairs
• Empty gloss detection — warns before export if glosses are incomplete

5. Customizable Categories

• Add/edit categories — create custom tag taxonomies for your project
• Dynamic hierarchy — add and remove categories, subcategories, and types on the fly
• Persistent storage — changes saved locally (no server required)
• Export categories — categorization system stored with your annotations

6. Data Management

• Auto-save — all work saved to browser local storage automatically
• Session restore — resume interrupted sessions with a single click
• Session export/import — download a complete session snapshot as JSON and import it on another device to continue seamlessly
• Clear data — full control to wipe local storage and start fresh
• Progress tracking — visual indicator of annotated vs. remaining phrases

---

Workflow

Quick Start

1. Load a file → Upload your CSV or Excel file with noun phrase data
2. Configure columns → Map which columns contain phrases, context, language, and code
3. Start annotating → Select phrases and tokens to tag
4. Add glosses → Provide word-level glossing and translations
5. Save phrases → Commit annotations; they are immediately available for tag suggestions
6. Export → Download lexicon and annotations in your preferred format

Example Annotation Flow

---

Default Tag Categories

The tool comes with a comprehensive default hierarchy:

Category	Subcategories	Types
NOUN	Animate, Inanimate	Object, Event
ADJ	Intersective, Non-intersective	Shape, Color, Material, Size, Age, Qualifier
ART	Definite, Indefinite	—
POSS	Genitive, PP-Genitive	—
NUM	Ordinal, Cardinal	—
DEM	Proximal, Distal	—
QUANT	Existential, Universal	—
RC	Restrictive, Non-restrictive	—
PP	(no subcategories)	—

Fully customizable — add, edit, or delete categories as needed for your project.

---

Export Formats

1. JSON Export

Complete annotation data in structured JSON:

• Phrases table (metadata + tag sequence)
• Tokens table (lexical info + glosses)
• Annotations table (tags, categories, token spans)

2. CSV Export

Three separate CSV files for relational database import:

• np_phrases_*.csv — phrase-level data
• np_tokens_*.csv — token-level data
• np_annotations_*.csv — annotation-level data

3. Excel (XLSX) Export

Single workbook with three sheets:

• phrases — one row per phrase
• tokens — one row per token
• annotations — one row per annotation

4. Lexicon Export

Standalone lexicon in JSON, CSV, or XLSX format:

• Lexeme ID, word form, language
• Meaning IDs with glosses
• Associated phrase IDs for each meaning

---

Browser Storage & Privacy

• No server — all data stored locally in browser localStorage
• No external API calls — fully offline-capable
• Auto-save — annotated data automatically saved with every action
• Session persistence — resume sessions across browser restarts
• Manual backup — export to JSON/CSV/XLSX for archival

Note: Clearing browser data will delete stored annotations. Export your work before clearing storage.