Datasetpublic

bhoomichowksey/consulting-sentiment-intelligence

🔍 Consulting Firm Review & Sentiment Intelligence Dataset Overview 6,200+ NLP-enriched reviews across 10 top consulting firms (MBB, Big 4, Tier 2), covering: 5,000 employee reviews — Glassdoor, AmbitionBox, Blind, Indeed, Comparably 1,200 portal reviews — Official firm client testimonials Each review includes VADER + TextBlob ensemble sentiment, 6-dimension aspect scores, NMF topic labels, and a firm-level Credibility Divergence Index. Files… See the full description on the dataset page: https://huggingface.co/datasets/bhoomichowksey/consulting-sentiment-intelligence.

sourceHugging Facemitupdated 3mo agoView on Hugging Face

0likes72downloads

Dataset Card

🔍 Consulting Firm Review & Sentiment Intelligence Dataset

Overview

6,200+ NLP-enriched reviews across 10 top consulting firms (MBB, Big 4, Tier 2), covering:

—5,000 employee reviews — Glassdoor, AmbitionBox, Blind, Indeed, Comparably
—1,200 portal reviews — Official firm client testimonials

Each review includes VADER + TextBlob ensemble sentiment, 6-dimension aspect scores, NMF topic labels, and a firm-level Credibility Divergence Index.

Files

File	Rows	Description
`employee_reviews_processed.csv`	5,000	Full NLP enrichment + aspects + topics
`portal_reviews_processed.csv`	1,200	Portal NLP scores
`firm_divergence_summary.csv`	10	Divergence Index per firm
`timeseries_enriched.csv`	719	Monthly aggregates + spike flags
`sentiment_forecast.csv`	60	6-month polynomial forecast
`yoy_drift.csv`	60	Year-over-year drift
`topic_distribution.csv`	70	NMF topic sentiment per firm
`aspect_monthly_trend.csv`	719	6-aspect monthly trends

Firms Covered

McKinsey & Company · Boston Consulting Group · Bain & Company · Deloitte Consulting · PwC Advisory · EY Consulting · KPMG Advisory · Accenture Strategy · Oliver Wyman · Roland Berger

Key Columns (employeereviewsprocessed.csv)

Column	Description
`ensemble_score`	Weighted sentiment (-1 to +1)
`ensemble_label`	positive / neutral / negative
`vader_compound`	VADER score (domain-tuned)
`tb_polarity`	TextBlob polarity
`tb_subjectivity`	TextBlob subjectivity
`confidence`	Model confidence
`aspect_work_life_balance`	WLB aspect score
`aspect_compensation`	Comp aspect score
`aspect_culture`	Culture aspect score
`aspect_leadership`	Leadership aspect score
`aspect_career_growth`	Growth aspect score
`aspect_work_quality`	Work quality aspect score
`topic_name`	NMF topic label
`bias_flags`	Quality flags

Usage

python

from datasets import load_dataset
ds = load_dataset("bhoomichowksey/consulting-sentiment-intelligence")
emp = ds["employee_reviews_processed"]

Citation

bibtex

@dataset{consulting_sentiment_2024,
  title={Consulting Firm Review & Sentiment Intelligence Dataset},
  author={Bhoomi Chowksey}, year={2024},
  publisher={Hugging Face}
}