---
language: en
license: mit
task_categories:
- text-classification
tags:
- structural-bottleneck
- stability-geometry
- reasoning
- clarus
- sios
size_categories:
- n<1K
pretty_name: Structural Bottleneck Classification v0.1
---

# What this dataset does

This dataset tests whether a model can detect structural bottlenecks.

The task is simple:

Given a scenario and a structural-bottleneck claim, predict whether the claim is supported.

# Core stability idea

A structural bottleneck is a constraint that limits system performance regardless of improvements elsewhere.

Typical bottlenecks include:

- single approval points
- single processing nodes
- unique dependencies
- centralized routing
- irreplaceable personnel
- constrained resources

Removing a bottleneck often increases system capacity more effectively than optimizing surrounding components.

# Prediction target

Binary label:

- 1 = a structural bottleneck is present
- 0 = a structural bottleneck is not present

# Row structure

Each row contains:

- scenario_id
- scenario_text
- claim
- label

# Files

- data/train.csv
- data/test.csv
- scorer.py
- README.md

# Evaluation

```bash
python scorer.py --predictions predictions.csv --truth data/test.csv

Structural Note

This dataset is intentionally small.

Its purpose is to test whether a model can identify limiting constraints embedded in system structure rather than transient operational issues.

License

MIT