Knowledge Systems / bulk-classifiersPrototype
Text & Image Classifiers
AI-assisted classification of text and images as a workstream for local bulk processing.
Problem
Initial situation
Large text and image collections need consistent assignments; ambiguous cases must not silently become binding categories.
Constraints
Boundary conditions
- Explicit taxonomies
- Traceable input references
- Uncertainty and review
- Protect original data
Solution
Solution
Specialized text and vision workers within AI Scaffolding use taxonomies, batch manifests and structured outputs. Text, geo and image classification use separate contracts; validators check categories, batch coverage and other criteria.
Architecture
Architecture
- Text classification
- Image classification
- Batch-oriented task pipelines
- Reviewable structured results
Result
Result
Individual and batch pathways and associated regression tests are present in the repository. The site review neither ran these tests nor evaluated models; throughput and classification accuracy remain unconfirmed.