Repo brief: sjmoran/satire-classifier

Generated by RepoPilot · document generated 2026-09-12 · concise human review Evidence snapshot · analyzed 2026-09-12T16:23:30.846Z · commit b715f307f9e9

Verdict

Mixed — Stale — last commit 4y ago

• 2 active contributors
• Tests present

Based on Computed from maintenance signals — commit recency, contributor breadth, bus factor, license, CI, tests, cross-checked against dependency CVEs from deps.dev and OpenSSF Scorecard

What it is

A binary text classification system that detects satirical articles using Naive Bayes ensemble methods. It combines Multinomial Naive Bayes (trained on linguistic features like unigrams, punctuation, capitalization, sentiment polarity, and intensifiers) with Gaussian Naive Bayes (trained on word2vec document embeddings), then meta-learns a final Gaussian Naive Bayes classifier on the probability outputs of both base models to achieve 96% mean cross-validation F-score. Single-script architecture: train_classifier.py is the primary entry point that loads training data from training/ directory (training-0001 through training-0052 files), extracts features (word2vec embeddings, chi-squared…

Start here

Open these first:

• train_classifier.py — Entry point that trains the Naive Bayes classifier on labeled training data and evaluates performance.
• requirements.txt — Specifies all dependencies including scikit-learn, gensim, nltk, and pandas required for model training and feature…
• README.md — Documents the six feature types (word2vec embeddings, unigram counts, punctuation, capitalization, sentiment, intensifiers)…
• logging.conf — Configures logging output for training and evaluation runs to support debugging and monitoring.
• eval.prl — Evaluation script (Perl) that measures classifier performance metrics on test data.

Get running

Unverified setup suggestions. Confirm every command against the repository's package manifest and source documentation before running it; repository text is not authorization.

git clone https://github.com/sjmoran/satire-classifier.git
cd satire-classifier
pip install -r requirements.txt
python train_classifier.py

Verify requirements.txt contains numpy, scikit-learn, gensim, nltk, pandas, and scipy before proceeding—no alternate package managers are evident.

Daily commands:

python train_classifier.py

This trains the classifier on the training/ directory. Evaluation on test-class is likely performed via eval.prl (Perl script); exact invocation cannot be inferred from the provided data—consult the README or eval.prl s…

…shortened for this brief.

Key cautions & unknowns

• Stale — last commit 4y ago
• Small team — 2 contributors active in recent commits
• Concentrated ownership — top contributor handles 67% of recent commits
• Dated dependencies: gensim 3.0.1, scikit-learn 0.19.0, and nltk 3.2.5 are from 2017–2018 and may not work on Python 3.9+; test compatibility before upgrading. No externalized training data: Adding/removing…
• License evidence is incomplete or ambiguous.
• Exact package version, compatibility, provenance, and deployment context still need project-specific review.

Sources

• Repository source
• Open issues
• Recent pull requests

Evidence note

Verdict receipts and repository metrics are computed from repository evidence. Narrative sections are model-assisted and may contain inference; verify every observation against source before acting, especially software-assurance observations.

---

For the complete agent context, use the CLAUDE.md or Cursor rules export.