Dataset Viewer
The dataset viewer is not available for this dataset.
Unexpected token '<', "<html> <h"... is not valid JSON

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Presentation link: https://aaudk-my.sharepoint.com/:v:/g/personal/ev52yu_student_aau_dk/IQBhbanv3sOmR4qpqKzNWJ8LASJjBYmnSK-GsC4yRPtShL8?nav=eyJyZWZlcnJhbEluZm8iOnsicmVmZXJyYWxBcHAiOiJPbmVEcml2ZUZvckJ1c2luZXNzIiwicmVmZXJyYWxBcHBQbGF0Zm9ybSI6IldlYiIsInJlZmVycmFsTW9kZSI6InZpZXciLCJyZWZlcnJhbFZpZXciOiJNeUZpbGVzTGlua0NvcHkifX0&e=spnj1M

🌱 Green Patent Classification using PatentSBERTa

Silver Supervision + Active Learning + Human-in-the-Loop Fine-Tuning

This repository contains a fine-tuned version of PatentSBERTa for binary classification of patent claims into:

  • 0 β€” Non-Green
  • 1 β€” Green Technology

We created a balanced dataset of 50,000 claims:

  • 25,000 Green (silver)
  • 25,000 Non-Green (silver)

Stratified split:

  • 70% Train
  • 15% Pool (for active learning)
  • 15% Evaluation

🧠 Baseline Model (Frozen Embeddings)

We computed claim embeddings using:

A Logistic Regression classifier was trained on frozen embeddings.

Silver Evaluation Performance

  • Accuracy β‰ˆ 77.5%
  • Balanced precision and recall across classes

This confirms that domain-specific embeddings capture sustainability semantics effectively even without fine-tuning.


πŸ” Active Learning Strategy

To improve labeling efficiency, we selected high-risk examples from the unlabeled pool using uncertainty sampling:

Where:

  • p_green is the model’s predicted probability for class 1.
  • Higher u indicates higher classification uncertainty.

The top 100 most uncertain examples were selected for expert review.


πŸ€– LLM β†’ Human HITL Workflow

LLM Used

qwen2.5:7b-instruct (via Ollama)

The LLM provided:

  • Suggested label (0/1)
  • Confidence level (low / medium / high)
  • Short rationale quoting the claim

Human reviewers assigned the final gold label.


πŸ”„ Human Overrides (Required Reporting)

Override defined as:

This measures disagreement between the LLM and human expert.

Example Overrides

  1. LLM 0 β†’ Human 1
    Claim: Conversion of biological material into energy.
    Reason: Biomass-based energy conversion qualifies as renewable energy under green technology criteria.

  2. LLM 0 β†’ Human 1
    Claim: Light concentrating optic for photovoltaic device.
    Reason: The invention directly improves solar energy generation, which is renewable and climate-mitigating.

Observed Pattern

The LLM tends to under-classify cases where renewable energy impact is implied but not explicitly framed as β€œclimate mitigation.”

This highlights the importance of human-in-the-loop review for borderline sustainability cases.


πŸš€ Final Fine-Tuning

We fine-tuned:

Training data:

  • Silver training split
    • 100 human-labeled gold examples

Training configuration:

  • Optimizer: AdamW
  • Learning rate: 2e-5
  • Epochs: 1
  • Max sequence length: 256
  • Batch size: 4

πŸ“ˆ Evaluation

The fine-tuned model was evaluated on:

  • Silver evaluation set
  • Gold HITL set

Fine-tuning improves performance especially on previously uncertain claims.


πŸ’» Usage Example

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

model_path = "your-model-path"

tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForSequenceClassification.from_pretrained(model_path)

text = "A system for converting biomass into renewable energy."

inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True, max_length=256)

with torch.no_grad():
    outputs = model(**inputs)

prediction = torch.argmax(outputs.logits, dim=-1).item()

print("Green" if prediction == 1 else "Non-Green")
Downloads last month
8