sea-rod/konkani-sentiment-analysis

Konkani Sentiment Analysis (Fine-tuned on Marathi Model)

This model is a sentiment analysis classifier that predicts one of three labels:

• positive
• negative
• neutral

It was fine-tuned from the existing l3cube-pune/MarathiSentiment model on additional data, with the goal of improving zero-shot performance on Konkani text.

---

Model Details

• Base Model: l3cube-pune/MarathiSentiment
• Languages: Marathi, Konkani
• Task: Sentiment Classification (positive, negative, neutral)
• Fine-tuning: We fine-tuned the model to improve cross-lingual transfer, especially for Konkani, a low-resource language closely related to Marathi.

---

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline

model_name = "sea-rod/konkani-sentiment-analysis"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSequenceClassification.from_pretrained(model_name)

sentiment_pipeline = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)

texts = [
    "मला हा चित्रपट खूप आवडला.",   # Marathi
    "हावनें खूप चांगले आसा."       # Konkani
]

predictions = sentiment_pipeline(texts)
print(predictions)

Example output:

[{'label': 'positive', 'score': 0.97},
 {'label': 'neutral', 'score': 0.65}]

---

Training

• Objective: Adapt a Marathi sentiment model to handle Konkani data using fine-tuning.
• Labels: positive, negative, neutral
• Motivation: Konkani has very limited labeled datasets. Leveraging a linguistically similar language (Marathi) allows for effective zero-shot and transfer learning.

---

Citation

If you use this model, please cite the following paper:

@inproceedings{m-ghosarwadkar-etal-2024-sentiment,
    title = "Sentiment Analysis for {K}onkani using Zero-Shot {M}arathi Trained Neural Network Model",
    author = "M. Ghosarwadkar, Rohit  and
      Rodrigues, Seamus Fred  and
      Bhagat, Pradnya  and
      Abranches, Alvyn  and
      Korkankar, Pratik Deelip  and
      Pawar, Jyoti",
    editor = "Lalitha Devi, Sobha  and
      Arora, Karunesh",
    booktitle = "Proceedings of the 21st International Conference on Natural Language Processing (ICON)",
    month = dec,
    year = "2024",
    address = "AU-KBC Research Centre, Chennai, India",
    publisher = "NLP Association of India (NLPAI)",
    url = "https://aclanthology.org/2024.icon-1.66/",
    pages = "569--575"
}

---

License

This model is licensed under the Creative Commons Attribution 4.0 International (CC-BY-4.0) license.

• Summary: Permits almost any use subject to providing credit and license notice. Commonly used for media assets, educational materials, and open-access publications. Not recommended for software.
• Full Text: Creative Commons Attribution 4.0 International License

When using this model, please provide proper attribution and include the license notice.