Documentation
← Back to Build

DATA & COMPUTATIONAL WORKFLOWS

Text Classification Lab

Test a deterministic baseline, review an AI classification suggestion, and inspect errors against human reference labels.

Research Preview

Outputs are AI-supported analytical suggestions. The researcher remains responsible for verifying and approving results for research use.

COMPUTATIONAL WORKBENCH

How this research workflow processes your data

MethodNaive Bayes baseline + separate AI inference
  1. 01Labeled DataText + human reference
  2. 02Label InspectionSize and class distribution
  3. 03PreprocessingArabic normalization and tokenization
  4. 04Train/Test SetupDeterministic stratified split
  5. 05Run Naive BayesDeterministic baseline
  6. 06Structured ResultsPredictions and confusion matrix
  7. 07EvaluationAccuracy and matches
  8. 08Error AnalysisSystem vs human reference
  9. 09ImprovementReview data, then re-run

Measured outputs are calculated by the stated method. AI interpretation, when offered, remains separate.

Baseline Classification

Enter each row as: text, Tab, human reference label. At least six records and three examples per label are required.

Pre-run data inspection

6Valid records
2Categories

Human-reference distribution: إيجابي: 3 · سلبي: 3

Deterministic baseline results

Enter valid labeled Arabic data with at least six records, two labels, and three examples per label.