Purpose
The Hyperon Pattern Miner is a core cognitive component for discovering meaningful structure inside the Atomspace. Its role is to identify frequent and surprising patterns, convert implicit graph activity into explicit symbolic knowledge, and provide higher-level reasoning components with reusable concepts, rules, and relations.
Why pattern mining matters for AGI
In the patternist view of mind, intelligence depends on the ability to recognize, store, create, and reuse patterns. A cognitive system operates in complex environments by compressing rich information into simpler and more useful representations.
Pattern recognition supports several major cognitive functions:
- Perception: identifying familiar or novel structures in data.
- Memory: storing recognized patterns for later use.
- Prediction: using temporal and structural patterns to anticipate useful actions.
- Reasoning: transforming discovered patterns into explicit concepts and rules.
For Hyperon, pattern mining helps move knowledge from implicit activity in the Atomspace to explicit symbolic structures that can support reasoning, learning, and cognitive synergy.
Hyperon Pattern Miner
The Hyperon Pattern Miner acts as an intuition and habit-formation mechanism for the system. It takes high-dimensional Atomspace data and extracts clean symbolic patterns that can be used by higher-level cognitive processes.
The miner has two main responsibilities:
- Frequent pattern mining: discovering patterns that occur often in the Hyperon Atomspace.
- Interestingness filtering: selecting the most informative patterns using surprisingness-based measures.
Frequent patterns alone may not always add useful new information. Therefore, the miner applies surprisingness algorithms to rank and filter results. A higher surprisingness score indicates that a pattern contributes more new information to the cognitive system.