CivicMap

Connecting Austin residents to local civic events that matter to them.

CivicMap is a web platform that centralizes the discovery of civic engagement opportunities in Austin, TX. Instead of checking multiple websites, users get a personalized feed of city council meetings, town halls, community forums, and political events based on their interests, preferred formats, and schedules.

---

Problem

• Fragmented Discovery: Civic events are scattered across city websites, Facebook, Eventbrite, and email lists
• Information Overload: Too many irrelevant events with no way to filter by topics you care about
• Lack of Context: Event descriptions are vague -- hard to tell if something is a routine meeting or a heated debate

---

Features

• Centralized Event Feed: All Austin civic events in one searchable, filterable feed
• Onboarding Questionnaire: 5-step preference setup (topics, tone, location, schedule, format) on first signup
• Personalized Recommendations: Events ranked by relevance to your selected interests
• Smart Categorization: Keyword-based auto-tagging of event topics (Housing, Environment, Education, Healthcare, Transportation, Public Safety, Economy, Social Justice)
• Tone Classification: Events labeled by tone (Informational, Contentious, Collaborative, Action-Oriented, etc.) so users know what to expect
• Event Detail Pages: View full event info with bookmark and register actions
• User Profiles: View and edit preferences, manage saved events
• Supabase Auth: Email/password signup and login with session management

---

Tech Stack

Layer	Technology
Frontend	Next.js 14 (App Router), React, TypeScript, Tailwind CSS
Database	Supabase (PostgreSQL)
Authentication	Supabase Auth (@supabase/ssr)
ML / Data Processing	Python, sentence-transformers, Supabase Python client
Hosting	AWS EC2

---

Database Schema

Key tables in Supabase:

• events -- id, title, description, event_date, location, event_url, categories, tone, sector, source, is_virtual, embedding
• categories -- id, name
• user_preferences -- user_id, tone, sector, schedule, format (all text arrays)
• user_category_prefs -- user_id, category_id
• user_interactions -- id, user_id, event_id, interaction_type, weight, created_at
• saved_events -- user_id, event_id, saved_at
• user_profiles -- id, location, created_at, updated_at

---

Machine Learning

Two Python scripts in the ml/ directory:

1. Keyword-Based Event Tagging (ml/tag_events.py)

   • Tags events with categories and tone based on keyword matching against event title and description
   • Reads from and writes back to the Supabase events table

2. Vector Embeddings (ml/generate_embeddings.py)

   • Generates 384-dimensional embeddings using sentence-transformers (all-MiniLM-L6-v2)
   • Combines title, description, and category into a single text for embedding
   • Stores embeddings in the embedding column for future similarity-based recommendations

---

Team

Build Team Lead: Ashi Sharma

Frontend

• Arav Bazar
• Shravan Venkat
• Khushi Bhalani

Backend

• Hana Alsayed
• Anshu Siripurapu
• Vishal Rajkumar

Machine Learning

• Nidhi Ilanthalavian
• Sasha Haltom
• Raina James

---

Contact

Project Lead: [email protected]