SocioInfer

socioinfer is a containerized microservice system that infers a Twitter user's interests by analyzing their profile and social graph. It combines on-demand data syncing, graph-based storage (Neo4j), and NLP-driven interest extraction.

Architecture

• twitter-interest-inference (FastAPI + Python)

  • Extracts interests from user and followings' bios using NLP.
  • Queries Neo4j directly for graph data.
  • Triggers syncing via NetworkSync.

• NetworkSync (Express + TypeScript)

  • Syncs a user's Twitter followings from the Twitter API into Neo4j.
  • Also supports and exposes APIs for mutual followings lookup and user graph storage.

• Neo4j (Graph Database)

  • Stores users and their following relationships as a graph.

Getting Started

Clone the repository (with submodules)

git clone --recurse-submodules https://github.com/pali101/SocioInfer.git
cd SocioInfer

Configure environment variables

Each service requires its own .env file.

For twitter interest inference:

cp ./twitter-interest-inference/.env.example ./twitter-interest-inference/.env

For network sync:

cp ./NetworkSync/.env.example ./NetworkSync/.env

Run the services

Both services are containerized and orchestrated via Docker Compose. Start everything using Docker Compose:

docker-compose up --build

This will launch all services, networked together and ready for API requests.

Structure

socioinfer/
├── NetworkSync/                   # Submodule: Twitter data sync
├── twitter-interest-inference/    # Submodule: interest inference API
├── docker-compose.yml
└── README.md

Tech Stack

• Backend: FastAPI (Python), Express (TypeScript)
• Database: Neo4j
• Containerization: Docker, Docker Compose
• NLP: Sentence Transformers

API Endpoints

Twitter Interest Inference (Port 8000)

• GET /interests/{username}: Returns inferred interests (optionally scored and model-selectable)

Network Sync (Port 4000)

• POST /api/sync: Syncs a user's Twitter followings into Neo4j
• POST /api/user/store: Stores a user in the graph
• GET /api/mutual?user1=&user2=: Checks mutual followings between two users