Table of Contents

Case studies

This folder contains numeric case studies that demonstrate Intentum's effectiveness in specific domains. Case study documents are in English only; they are not duplicated under docs/en or docs/tr.

  • Greenwashing detection metrics — labeled dataset, transformation rule, accuracy/F1 summary. Three evaluation paths: (1) synthetic: GreenwashingCaseStudy_ComputesAccuracyAndF1 (19 examples); (2) public HTML: run ./scripts/download-greenwashing-sources.sh, then GreenwashingCaseStudy_OnDownloadedHtml_ComputesAccuracyAndF1 (ClientEarth); (3) Mendeley Excel: unpack dataset to downloaded/DataGreenwash/, then GreenwashingCaseStudy_OnMendeleyExcel_ComputesAccuracyAndF1. Sources: greenwashing-sources.md, greenwashing-labeled-sources.csv (53 rows).
  • Cross-LLM consistency — optional: how to compare confidence/intent across embedding providers (OpenAI, Azure, Gemini, etc.).
  • Load test: infer endpoint — k6 script and result summary for Sample.Blazor POST /api/intent/infer.
  • Benchmark results — BenchmarkDotNet results (ToVector, Infer, PolicyEngine). Generated by ./scripts/run-benchmarks.sh.
  • Blazor five demo specs — Detaylı spesifikasyonlar: Finans Dolandırıcılığı, ATO, İçeriden Tehdit, Sıfır-Gün, API/Web Trafiği demo senaryolarının akışı, event listeleri, görsel metinleri ve politika eşikleri (Blazor örnek projesi implementasyon referansı).
  • Intent Layer five demos — Intent Layer 5 devrimsel demo: Project Pulse, Customer Journey, Moderation, Adaptive Tutor, Digital Twin. Status tabloları, event katalogları, varyant eşlemesi (A/B/C/D), politika kuralları ve API özeti.