Wotan – A Natural Language Generation System Overview

Wotan (NLG System)

Overview

Wotan is a natural language generation (NLG) system designed to transform structured or semi-structured data into human-readable text. Developed as part of ongoing research in computational linguistics and AI language automation, Wotan demonstrates how symbolic rules and linguistic models can work together to produce clear, fluent, and context-aware text.

It is documented within the NLG Systems Wiki — a collaborative knowledge base that catalogs architectures, inputs, outputs, and applications of text generation systems.

Architecture and Design

The Wotan architecture follows a modular NLG pipeline, commonly divided into:

1. Content Determination – selecting relevant facts or data to communicate.

2. Document and Sentence Planning – structuring the narrative and choosing appropriate sentence patterns.

3. Lexicalisation – mapping concepts to suitable words or phrases.

4. Surface Realisation – producing grammatically correct sentences from linguistic structures.

Unlike modern neural generators, Wotan relies on symbolic rules and templates, giving developers fine-grained control over the output. This design makes it highly interpretable and adaptable for domain-specific applications such as technical reporting, knowledge summarisation, or educational content.

Input and Output

Wotan typically takes structured or semi-structured data as input — such as database entries, logical propositions, or ontology triples — and outputs coherent text in natural language.

It supports multiple generation strategies, including:

• Rule-based NLG, where output is determined by explicit linguistic patterns.

• Template-based realisation, for predictable or domain-constrained contexts.

• Hybrid configurations, allowing partial statistical or heuristic decisions.

Applications

Wotan is designed for use in environments where accurate, consistent text is needed:

• Business reporting and data summaries

• Software documentation generation

• Education and training tools

• AI-assisted writing for structured datasets

By automatically turning facts into readable narratives, Wotan reduces manual authoring time and ensures consistency in output style.

Research and Development Context

Within the broader NLG ecosystem, Wotan serves as a reference point for rule-driven text generation. It highlights the advantages of explicit knowledge representation — transparency, control, and linguistic precision — which are sometimes lost in neural black-box models. Its documentation on NLG-Wiki includes architecture diagrams, linguistic modules, and comparison with other systems

FAQ – Wotan System

What is Wotan used for?
Wotan is used to automatically generate readable text from structured data sources such as databases, knowledge bases, or domain ontologies.

Is Wotan rule-based or AI-driven?
It is primarily rule-based, though it can integrate heuristic or hybrid components for flexibility.

Can Wotan be used in multiple languages?
Depending on configuration, Wotan can support multilingual output through modular grammar resources.

Where can I learn more about Wotan?
You can explore full system details on its entry page at nlg-wiki.org/systems/Wotan.