SubjECTiveQA-CLEAR Model

Model Name: SubjECTiveQA-CLEAR

Model Type: Text Classification

Language: English

Base Model: google-bert/bert-base-uncased

Dataset Used for Training: gtfintechlab/SubjECTive-QA

Model Overview

SubjECTiveQA-CLEAR is a fine-tuned BERT-based model designed to classify text data according to the 'CLEAR' attribute. The 'CLEAR' attribute is one of several subjective attributes annotated in the SubjECTive-QA dataset, which focuses on subjective question-answer pairs in financial contexts.

Intended Use

This model is intended for researchers and practitioners working on subjective text classification, particularly within financial domains. It is specifically designed to assess the 'CLEAR' attribute in question-answer pairs, aiding in the analysis of subjective content in financial communications.

How to Use

To utilize this model, you can load it using the Hugging Face transformers library:

from transformers import pipeline, AutoTokenizer, AutoModelForSequenceClassification, AutoConfig

# Load the tokenizer, model, and configuration
tokenizer = AutoTokenizer.from_pretrained("gtfintechlab/SubjECTiveQA-CLEAR", do_lower_case=True, do_basic_tokenize=True)
model = AutoModelForSequenceClassification.from_pretrained("gtfintechlab/SubjECTiveQA-CLEAR", num_labels=3)
config = AutoConfig.from_pretrained("gtfintechlab/SubjECTiveQA-CLEAR")

# Initialize the text classification pipeline
classifier = pipeline('text-classification', model=model, tokenizer=tokenizer, config=config, framework="pt")

# Classify the 'CLEAR' attribute in your question-answer pairs
qa_pairs = [
    "Question: What are your company's projections for the next quarter? Answer: We anticipate a 10% increase in revenue due to the launch of our new product line.",
    "Question: Can you explain the recent decline in stock prices? Answer: Market fluctuations are normal, and we are confident in our long-term strategy."
]
results = classifier(qa_pairs, batch_size=128, truncation="only_first")

print(results)

Label Interpretation

LABEL_0: Negatively Demonstrative of 'CLEAR' (0)
Indicates that the response lacks clarity.
LABEL_1: Neutral Demonstration of 'CLEAR' (1)
Indicates that the response has an average level of clarity.
LABEL_2: Positively Demonstrative of 'CLEAR' (2)
Indicates that the response is clear and transparent.

Training Data

The model was trained on the SubjECTive-QA dataset, which comprises question-answer pairs from financial contexts, annotated with various subjective attributes, including 'CLEAR'. The dataset is divided into training, validation, and test sets, facilitating robust model training and evaluation.

Citation

If you use this model in your research, please cite the SubjECTive-QA dataset:

@article{SubjECTiveQA,
  title={SubjECTive-QA: Measuring Subjectivity in Earnings Call Transcripts’ QA Through Six-Dimensional Feature Analysis},
  author={Huzaifa Pardawala, Siddhant Sukhani, Agam Shah, Veer Kejriwal, Abhishek Pillai, Rohan Bhasin, Andrew DiBiasio, Tarun Mandapati, Dhruv Adha, Sudheer Chava},
  journal={arXiv preprint arXiv:2410.20651},
  year={2024}
}

For more details, refer to the SubjECTive-QA dataset documentation.

Contact

For any SubjECTive-QA related issues and questions, please contact:

Huzaifa Pardawala: huzaifahp7[at]gatech[dot]edu
Siddhant Sukhani: ssukhani3[at]gatech[dot]edu
Agam Shah: ashah482[at]gatech[dot]edu

gtfintechlab
/

SubjECTiveQA-CLEAR