NLP and network analytics platform for measuring professional brand performance on LinkedIn.

Built as a portfolio project demonstrating ML/AI engineering capabilities: sentiment analysis, topic modeling, network graph analytics, and interactive visualization.

The LinkedIn Brand Analyzer transforms your LinkedIn data export into actionable insights about your professional brand. It answers questions like:

• What topics drive the most engagement with my content?
• How does my posting sentiment correlate with audience response?
• Who are the high-value connections in my network (recruiters, decision-makers)?
• How do I compare to industry benchmarks?

Sentiment Analysis	Topic Distribution	Network Graph

Engagement ROI	Target Companies

LinkedIn Data Export → Ingestion Pipeline → NLP Analysis → Network Analytics → Dashboard
                              ↓                   ↓                ↓
                        Cleaned CSVs         Sentiment         Community
                                             Topics            Centrality
                                             Entities          PageRank

• VADER lexicon-based sentiment scoring
• Sentiment-engagement correlation analysis
• Temporal sentiment trends

• Configurable keyword-based topic matching
• Support for BERTopic/LDA topic modeling
• Engagement metrics by topic category

• Professional network graph construction
• Community detection using Louvain algorithm
• Centrality metrics (degree, betweenness, PageRank)
• Interactive network visualization

• Title-based recruiter identification
• Tier-1 company classification
• Engagement scoring weighted by connection value

• Configurable influencer benchmark data
• Percentile ranking across key metrics
• Industry-specific comparisons

• Python 3.9+
• LinkedIn data export (Settings → Data Privacy → Get a copy of your data)

git clone https://github.com/rosalinatorres/linkedin-brand-analyzer.git
cd linkedin-brand-analyzer
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt
python -m spacy download en_core_web_sm

1. Request your data export from LinkedIn (Settings → Data Privacy → Get a copy of your data)
2. Wait for email (24-72 hours)
3. Extract to data/raw/linkedin_export/

streamlit run app/streamlit_app.py

linkedin-brand-analyzer/
├── config/
│   └── settings.yaml           # Configuration & influencer benchmarks
├── data/
│   ├── raw/                    # LinkedIn export data
│   ├── processed/              # Cleaned datasets
│   └── benchmarks/             # Industry comparison data
├── src/
│   ├── ingestion/              # Data loaders & parsers
│   │   └── linkedin_loader.py
│   ├── nlp/                    # NLP analysis modules
│   │   └── analyzer.py         # Sentiment, topics, NER
│   ├── network/                # Graph analytics
│   │   └── graph_builder.py    # NetworkX graphs, community detection
│   ├── engagement/             # Scoring & detection
│   │   └── scorer.py           # Recruiter detection, engagement metrics
│   └── visualization/          # Chart generation
├── app/
│   └── streamlit_app.py        # Interactive dashboard
├── notebooks/                  # Jupyter exploration & analysis
├── tests/                      # Unit tests
├── images/                     # Generated visualizations
├── requirements.txt
└── run.py                      # CLI entry point

• NLP: VADER, TextBlob, SpaCy, BERTopic
• Network Analysis: NetworkX, python-louvain, PyVis
• Visualization: Plotly, Streamlit, Matplotlib
• Data Processing: Pandas, NumPy

# Launch interactive dashboard
python run.py --dashboard

# Process LinkedIn data export
python run.py --process /path/to/linkedin/export

# Run full analysis pipeline
python run.py --analyze

from src.nlp.analyzer import SentimentAnalyzer, KeywordTopicClassifier
from src.network.graph_builder import NetworkBuilder, NetworkAnalyzer
from src.engagement.scorer import RecruiterDetector

# Sentiment analysis
analyzer = SentimentAnalyzer(method='vader')
results = analyzer.analyze_batch(posts['content'].tolist())

# Network analysis
builder = NetworkBuilder()
graph = builder.build_from_connections(connections_df)
network_analyzer = NetworkAnalyzer(graph)
communities = network_analyzer.detect_communities()

# Recruiter detection
detector = RecruiterDetector()
recruiters = detector.detect_batch(connections_df)

This project demonstrates proficiency in:

1. NLP Engineering: Sentiment analysis, topic modeling, named entity recognition
2. Graph Analytics: Network construction, community detection, centrality measures
3. Data Engineering: ETL pipelines, data quality, schema design
4. ML Engineering: Classification, clustering, feature engineering
5. Full-Stack Development: Streamlit dashboards, interactive visualization

• BERTopic integration for advanced topic modeling
• Time-series forecasting for engagement prediction
• A/B testing framework for content optimization
• LinkedIn API integration for real-time data
• Export to Memory Brain for cross-system intelligence

Rosalina Torres
MS Data Analytics Engineering, Northeastern University
LinkedIn | GitHub

MIT License - see LICENSE for details.