kurry/earnings_transcripts_db

Earnings Call Transcripts Dataset

Overview

This dataset contains a comprehensive collection of earnings call transcripts for major healthcare and pharmaceutical companies. These transcripts capture quarterly discussions between company executives and financial analysts, providing unique insights into company performance, strategic directions, and market outlooks.

These high-quality primary source materials offer valuable text data for financial sentiment analysis, market intelligence research, and various NLP applications.

Dataset Specifications

Attribute	Value
Total Transcripts	761
Companies	3 major healthcare/pharma corporations
Time Period	2007-01-25 to 2025-03-24
Average Transcript	~58,229 characters
Format	Hugging Face Dataset (Arrow)
Language	English

Companies Included

• UNH - UnitedHealth Group Incorporated
• LLY - Eli Lilly and Company
• NVO - Novo Nordisk A/S

Data Structure

Each record contains the following fields:

Field	Description
ticker	Stock ticker symbol (e.g., "UNH")
company_name	Full legal company name
companyid	Unique company identifier
announcedate	Date of the earnings announcement
headline	Title of the earnings call (typically including quarter and year)
keydevid	Unique key development identifier
transcriptid	Unique transcript identifier
transcript_text	Complete transcript text including executive presentations and Q&A

Applications

This dataset is particularly valuable for:

• Financial NLP Research:

  • Sentiment analysis of executive communications
  • Topic modeling across earnings calls
  • Named entity recognition in financial contexts
  • Question-answering systems training

• Financial Market Analysis:

  • Extracting forward-looking statements
  • Analyzing management tone and confidence
  • Tracking clinical trial updates and product developments
  • Studying analyst concerns through Q&A patterns

• Healthcare/Pharma Intelligence:

  • Monitoring competitive positioning
  • Tracking industry terminology evolution
  • Analyzing discussion of regulatory challenges
  • Identifying emerging market trends

Usage Example

from datasets import load_from_disk

# Load the dataset
dataset = load_from_disk("earnings_transcripts_db")

# Basic info
print(f"Dataset contains {len(dataset)} transcripts")
print(f"Companies: {sorted(dataset.unique('ticker'))}")

# Sample analysis
import random
random_idx = random.randint(0, len(dataset)-1)
sample = dataset[random_idx]
print(f"Sample from {sample['company_name']} ({sample['ticker']}), {sample['announcedate']}")
print(f"Transcript length: {len(sample['transcript_text'])} chars")
print(f"First 500 chars: {sample['transcript_text'][:500]}...")

Citation

If you use this dataset in your research, please cite it as:

@dataset{earnings_transcripts_2025,
  author    = {Tran, Kurry},
  title     = {Healthcare and Pharmaceutical Earnings Call Transcripts},
  year      = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/...}}
}

License

This dataset is available for academic and research purposes. Usage for commercial applications may require additional licensing from the original content owners.

Limitations

• While extensive, this dataset covers a subset of healthcare and pharmaceutical companies.
• Some earnings calls may have missing segments or formatting irregularities.
• Multiple versions of the same transcript (preliminary, edited, etc.) may be present.
• Users should perform their own verification for critical research applications.