Over the last three years, I have worked with Christopher Alexander’s A Pattern Language and Robert E. Horn’s Information Mapping. My goal was to create reusable and machine-readable technical documentation for both human readers and generative AI.

Alexander warned against modularity because it risks losing the relationships inherent in good design. Despite this, modularity became necessary. Technical documentation requires clarity, consistency, and precision. Modular structures achieve these qualities.

To address the tension between Alexander’s organic approach and Horn’s structured methodology, I adopted Horn’s Information Mapping. This method identifies clear information types: procedures, concepts, principles, and facts. Integrating Alexander’s patterns with Horn’s modular approach provided clarity and usability.

Patterns offered a method to formalize repeated content structures. Patterns help capture and reuse documentation. Combined with generative AI and Noam Chomsky’s Generative Grammar, patterns guide the creation of consistent content.

Based on this approach, I developed the Pattern Language Miner. The tool extracts and categorizes recurring content patterns from Markdown, HTML, and plain text. It uses Natural Langauge Processing (NLP) techniques and semantic clustering. The Pattern Language Miner builds a knowledge base of reusable documentation patterns.

Currently, the tool focuses on content “chunks.” My plan includes adding further abstraction layers. These layers will help organize text blocks more effectively.

The Pattern Language Miner represents progress toward structured modularity and pattern-based documentation. It facilitates clear, consistent technical writing and supports integration with AI tools. I will continue developing this method for technical writing and knowledge engineering.