📊 YouTube Comment Intelligence

Semantic sentiment analysis of YouTube comments — powered by RoBERTa, Zero-Shot Classification, and Gemini 2.0 Flash expert reporting.

---

✨ Features

Feature	Details
Comment Fetching	YouTube Data API v3 — up to 500 comments per run
Sentiment Analysis	RoBERTa (cardiffnlp/twitter-roberta-base-sentiment-latest) — 99%+ accuracy
Semantic Clustering	Zero-Shot Classification (BART MNLI) into 3 audience segments
Interactive Dashboard	Plotly charts: sentiment pie, cluster bar, cross-matrix, activity timeline
🤖 Expert AI Report	Gemini 2.0 Flash — dynamic analyst role + 3 business opportunity recommendations
Export	Full CSV + summary CSV + expert report .md download
Demo Mode	Run with built-in sample data — no API keys required

---

🧠 How the Expert Report Works

After sentiment analysis is complete, the app sends a structured data digest to Gemini 2.0 Flash, which assumes a dynamic expert role based on the dominant comment cluster:

Dominant Cluster	Analyst Role Assigned
🪖 Military/Hawkish	Senior Geopolitical & Defense Policy Analyst
💰 Economic/Domestic	Senior Economic Analyst (Consumer Sentiment)
🗳️ Electoral/Party	Senior Political Communications Strategist
Other	Senior Social Media Intelligence Analyst

The report includes:

1. Executive Summary — dominant mood and key signal
2. Cluster-by-Cluster Analysis — tone, themes, notable examples
3. Audience Intent & Behavior Signals — what this audience is ready to do
4. Risk & Opportunity Signals — narratives gaining or losing momentum
5. Top 3 Business Opportunities — specific, data-grounded recommendations

---

🚀 Deploy on Streamlit Cloud (Free)

1. Fork this repo to your GitHub account
2. Go to share.streamlit.io
3. Connect your GitHub account and select this repo
4. Set app.py as the main file → click Deploy

⏱️ First run: ~5 min to download NLP models (~2 GB). Subsequent runs use cache.

---

🔑 API Keys Required

YouTube Data API (for live comment fetching)

1. Go to console.cloud.google.com
2. Create a new project
3. Enable YouTube Data API v3
4. Create credentials → API Key
5. Paste into the sidebar

Google Gemini API (for expert report — free tier)

1. Go to aistudio.google.com/app/apikey
2. Sign in with your Google account
3. Click Create API Key
4. Paste into the sidebar under Expert Analysis

Free tier limits: 15 requests/min · 1,500 requests/day — more than enough for analysis sessions.

---

💻 Local Development

git clone https://github.com/GreatFooliao111/youtube-comment-intelligence.git
cd youtube-comment-intelligence
pip install -r requirements.txt
streamlit run app.py

---

🏗️ Architecture

YouTube Data API v3
        │
        ▼
  Comment Fetcher
  (up to 500 comments)
        │
        ▼
   Text Cleaner
   (URLs, mentions, noise removal)
        │
        ├──────────────────────────┐
        ▼                          ▼
RoBERTa Sentiment         BART Zero-Shot
(Positive/Negative/        Cluster Classifier
 Neutral + score)          (Military / Economic
                            / Electoral)
        │                          │
        └──────────┬───────────────┘
                   ▼
        Streamlit Dashboard
        (Charts · Tables · CSV Export)
                   │
                   ▼
        Gemini 2.0 Flash
        Expert Analyst Report
        (Dynamic Role · Business Opportunities)

---

🛠️ Tech Stack

Layer	Technology
UI & Dashboard	Streamlit
Sentiment Model	cardiffnlp/twitter-roberta-base-sentiment-latest (HuggingFace)
Cluster Classifier	facebook/bart-large-mnli (Zero-Shot, HuggingFace)
Expert Report	Google Gemini 2.0 Flash (google-generativeai)
Charts	Plotly Express
Data	YouTube Data API v3 (google-api-python-client)

---

📄 License

MIT — free to use, modify, and deploy.