🌱 SeedGuard AI

Mapping the GMO Narrative Landscape for True Agricultural Empowerment & Sovereignty

---

📘 Overview

SeedGuard AI is a Natural Language Processing (NLP) and Machine Learning (ML) project that explores the global conversation surrounding genetically modified organisms (GMOs) and seed sovereignty across Africa. By collecting thousands of publicly available discussions from X App (Twitter) and Reddit, our system identifies, analyzes, and maps emerging narratives, emotional tones, and stakeholder perspectives shaping Africa’s agricultural future.

---

🎯 Objectives

• Collect and preprocess text data from online platforms such as X (Twitter) or QUORA (alternatively) and Reddit using targeted keywords (e.g., African seed sovereignty, agroecology, genetically modified maize, genetically modified food, GMO in Africa, etc.).
• Apply advanced NLP models (DistilBERT, BERTopic) to automatically cluster, categorize, and analyze public opinions.
• Map connected topics, sentiment shifts, and emotional distributions across African regions.
• Support evidence-based discussions and policy decisions on agricultural sovereignty and biotechnology.

---

🧩 Project Workflow

1️⃣ Data Collection & Preprocessing

• Tools: Apify, PRAW, BeautifulSoup, requests
• Cleaning: Remove stopwords, URLs, emojis, duplicates.
• Languages: English (focus), others optional via translation models.
• Output: Structured CSV/JSON for modelling.

2️⃣ NLP Modeling & Clustering

• Embedding Model: DistilBERT via sentence-transformers
• Topic Modelling: BERTopic with UMAP + HDBSCAN
• Sentiment Analysis: VADER, TextBlob
• Classification: Predict stakeholder groups (e.g., farmers, companies, NGOs).

3️⃣ Visualization & Insights

• Tools: Plotly, Seaborn, Matplotlib, Folium
• Dashboards: Topic frequency charts, sentiment maps, stakeholder influence graphs.
• Platforms: Streamlit or Dash for deployment.

---

🏗️ Folder Structure

C9-team-seedguard/
│
├── data/
│                 # Unprocessed data (excluded from Git)
│   
│   
│
├── notebooks/              # Jupyter notebooks for exploration
│
│   ├── preprocessed.ipynb
│   ├── classification.ipynb
│   ├── topic_modeling.ipynb
│   ├── sentiment_analysis.ipynb
│   └── visualization.ipynb
│
├── results/                # Generated insights, plots, figures
├── requirements_data.txt   # Dependencies for data & visualization
├── requirements_model.txt  # Dependencies for model training
├── .gitignore
└── README.md

---

⚙️ Installation & Setup

1️⃣ Clone the Repository

git clone https://github.com/AISaturdaysLagos/C9-team-seedguard.git
cd C9-team-seedguard

2️⃣ Create a Virtual Environment

python -m venv venv
venv\Scripts\activate   # For Windows
# or
source venv/bin/activate  # For Mac/Linux

3️⃣ Install Dependencies

For data processing only:

pip install -r requirements_data.txt

For model training:

pip install -r requirements_model.txt

4️⃣ Setup API Keys

Create a .env file and add your credentials:

TWITTER_BEARER_TOKEN=your_twitter_api_key
REDDIT_CLIENT_ID=your_reddit_client_id
REDDIT_SECRET=your_reddit_secret

---

🧠 Key Technologies

• Data Collection: apify, PRAW, BeautifulSoup
• NLP & Modeling: DistilBERT, BERTopic, Transformers, HDBSCAN, UMAP
• Visualization: Plotly, Folium, Seaborn, Dash, Streamlit
• Languages: Python, Jupyter Notebook

---

📊 Expected Deliverables

• Clean dataset of GMO-related discussions across Africa.
• NLP model capable of identifying key narratives and stakeholder categories.
• Interactive visualization dashboard for exploring topic and sentiment trends.
• Technical report summarizing findings, insights, and recommendations.

---

👥 Team Roles

Role	Name	Responsibilities
Tech Lead	Ismail O. Daud	Project structure, workflow, integration, model supervision
Data Engineers	Agoro Zeenat	Data collection, cleaning, and preprocessing
ML Engineers	Elijah Aremu, Tijani O.	Topic modeling, embeddings, classification
Visualization Analysts	All team members	Charts, dashboards, storytelling
Documentation Lead	Ismail O. Daud	Reports, README, presentation materials

---

🚀 Future Extensions

• Add multilingual sentiment analysis for Francophone and Arabic-speaking Africa.
• Integrate real-time social media tracking APIs.
• Develop a Streamlit dashboard with live trend updates.
• Deploy models to Hugging Face or Google Cloud for public access.

---

📜 License

This project is developed under the AI Saturdays Lagos Cohort 9 program for machine learning and research purpose.

---

💬 Contact

For inquiries or collaborations:

• 📧 [email protected]