---
license: mit
language:
  - en
tags:
  - document-classification
  - scientific-papers
  - ai-detection
  - toxicity-detection
  - model2vec
  - pubverse
  - publication-screening
  - quality-control
library_name: model2vec
pipeline_tag: text-classification
thumbnail: PubGuard.png
---

<div align="center">
  <img src="PubGuard.png" alt="PubGuard Logo" width="400"/>
</div>

# PubGuard — Multi-Head Scientific Publication Gatekeeper

## Model Description

PubGuard is a lightweight, CPU-optimized document classifier that screens PDF text to determine whether it represents a genuine scientific publication. It runs as **Step 0** in the PubVerse + 42DeepThought pipeline, rejecting non-publications (posters, abstracts, flyers, invoices) before expensive downstream processing (VLM feature extraction, graph construction, GNN scoring).

Three classification heads provide a multi-dimensional screening verdict:

1. **Document type** — Is this a paper, poster, abstract, or junk?
2. **AI detection** — Was this written by a human or generated by an LLM?
3. **Toxicity** — Does this contain toxic or offensive content?

Developed by Jamey O'Neill at the California Medical Innovations Institute (CalMI²).

## Architecture

Three linear classification heads on frozen [model2vec](https://github.com/MinishLab/model2vec) (potion-base-32M) embeddings:

```
┌─────────────┐
│  PDF text    │
└──────┬──────┘
       │
┌──────▼──────┐     ┌───────────────────┐
│  clean_text │────►│  model2vec encode  │──► emb ∈ R^512
└─────────────┘     └───────────────────┘
                            │
          ┌─────────────────┼─────────────────┐
          ▼                 ▼                  ▼
┌─────────────────┐ ┌──────────────┐ ┌──────────────┐
│ doc_type head    │ │ ai_detect    │ │ toxicity     │
│ [emb + 14 feats] │ │ head         │ │ head         │
│ → softmax(4)     │ │ → softmax(2) │ │ → softmax(2) │
└─────────────────┘ └──────────────┘ └──────────────┘
```

Each head is a single linear layer stored as a numpy `.npz` file (8–12 KB). Inference is pure numpy — no torch needed at prediction time.

The `doc_type` head additionally receives 14 structural features (section headings present, citation density, sentence length, etc.) concatenated with the embedding — these act as strong Bayesian priors.

## Performance

| Head | Classes | Accuracy | F1 |
|------|---------|----------|-----|
| **doc_type** | 4 | **99.7%** | 0.997 |
| **ai_detect** | 2 | 83.4% | 0.834 |
| **toxicity** | 2 | 84.7% | 0.847 |

### doc_type Breakdown

| Class | Precision | Recall | F1 |
|-------|-----------|--------|-----|
| scientific_paper | 1.000 | 1.000 | 1.000 |
| poster | 0.989 | 0.974 | 0.981 |
| abstract_only | 0.997 | 0.997 | 0.997 |
| junk | 0.993 | 0.998 | 0.996 |

### Throughput

- **302 docs/sec** single-document, **568 docs/sec** batched (CPU only)
- **3.3ms** per PDF screening — negligible pipeline overhead
- No GPU required

## Gate Logic

Only `scientific_paper` passes the gate. Everything else — posters, standalone abstracts, junk — is blocked. The PubVerse pipeline processes **publications only**.

```
scientific_paper  →  ✅ PASS
poster            →  ❌ BLOCKED  (classified, but not a publication)
abstract_only     →  ❌ BLOCKED
junk              →  ❌ BLOCKED
```

AI detection and toxicity are **informational by default** — reported but not blocking.

## Usage

### Python API

```python
from pubguard import PubGuard

guard = PubGuard()
guard.initialize()

verdict = guard.screen("Introduction: We present a novel deep learning approach...")
print(verdict)
# {
#   'doc_type': {'label': 'scientific_paper', 'score': 0.994},
#   'ai_generated': {'label': 'human', 'score': 0.875},
#   'toxicity': {'label': 'clean', 'score': 0.999},
#   'pass': True
# }
```

### Pipeline Integration (bash)

```bash
# Step 0 in run_pubverse_pipeline.sh:
PDF_TEXT=$(python3 -c "import fitz; d=fitz.open('$pdf'); print(' '.join(p.get_text() for p in d)[:8000])")
PUBGUARD_CODE=$(echo "$PDF_TEXT" | python3 pub_check/scripts/pubguard_gate.py 2>/dev/null)
# exit 0 = pass, exit 1 = reject
```

### Installation

```bash
pip install git+https://github.com/jimnoneill/pubguard.git
```

With training dependencies:

```bash
pip install "pubguard[train] @ git+https://github.com/jimnoneill/pubguard.git"
```

## Training Data

Trained on real datasets from HuggingFace — **zero synthetic junk data**:

| Head | Sources | Samples |
|------|---------|---------|
| **doc_type** | armanc/scientific_papers, gfissore/arxiv-abstracts-2021, ag_news, [poster-sentry-training-data](https://huggingface.co/datasets/fairdataihub/poster-sentry-training-data) | ~55K |
| **ai_detect** | liamdugan/raid (abstracts), NicolaiSivesind/ChatGPT-Research-Abstracts | ~30K |
| **toxicity** | google/civil_comments, skg/toxigen-data | ~30K |

The poster class uses real scientific poster text from the [posters.science](https://posters.science) corpus (28K+ verified posters from Zenodo & Figshare), extracted by [PosterSentry](https://huggingface.co/fairdataihub/poster-sentry).

### Training

```bash
python scripts/train_pubguard.py --data-dir ./pubguard_data --n-per-class 15000
```

Training completes in ~1 minute on CPU. No GPU needed.

## Model Specifications

| Attribute | Value |
|-----------|-------|
| Embedding backbone | minishlab/potion-base-32M (model2vec StaticModel) |
| Embedding dimension | 512 |
| Structural features | 14 (doc_type head only) |
| Classifier | LogisticRegression (sklearn) per head |
| Head file sizes | 5–9 KB each (.npz) |
| Total model size | ~125 MB (embedding) + 20 KB (heads) |
| Precision | float32 |
| GPU required | No (CPU-only) |
| License | MIT |

## Citation

```bibtex
@software{pubguard_2026,
  title = {PubGuard: Multi-Head Scientific Publication Gatekeeper},
  author = {O'Neill, James},
  year = {2026},
  url = {https://huggingface.co/jimnoneill/pubguard-classifier},
  note = {Part of the PubVerse + 42DeepThought pipeline}
}
```

## License

This model is released under the [MIT License](https://opensource.org/licenses/MIT).

## Acknowledgments

- California Medical Innovations Institute (CalMI²)
- [MinishLab](https://github.com/MinishLab) for the model2vec embedding backbone
- [FAIR Data Innovations Hub](https://fairdataihub.org/) for the [PosterSentry](https://huggingface.co/fairdataihub/poster-sentry) training data
- HuggingFace for model hosting infrastructure