Digital records rankings coming soon as more data is collected across all boroughs.
Digital records rankings coming soon as more data is collected across all boroughs.
Analysis of demographic trends across Brooklyn Community Board appointees from 2022 to 2025. Data sourced from Brooklyn Borough President's annual demographic reports.
This dashboard was created to analyze and visualize NYC Community Board needs statements across multiple fiscal years. The process combines web scraping, PDF text extraction, and AI-powered analysis to identify policy trends and priorities.
Data Collection Pipeline
- PDF Scraping: Automated collection of Department of City Planning (DCP) needs statements for Brooklyn Community Boards (FY2024-2026)
- Text Extraction: Using
pdfplumberto extract text from 52 Brooklyn PDF documents - Database Storage: SQLite database (
dns_data.db) storing extracted text with metadata - Year-Over-Year Comparison: GPT-4 analysis comparing how priorities evolved across 3 fiscal years
- Theme Extraction: Secondary LLM pipeline to automatically identify 4-6 major themes per community board
LLM Analysis System Prompts
The AI-powered analysis uses carefully designed prompts to ensure accurate theme extraction and comparison:
Year-Over-Year Comparison Prompt
Instructs GPT-4 to compare THREE fiscal years of needs statements, focusing on:
- Policy Changes: What priorities appeared, disappeared, or shifted
- Agency Response Evolution: How city agency tone/engagement changed
- Emphasis Shifts: Which themes grew stronger or faded
- Structural Changes: Document format and organization differences
View Full Prompt →
CRITICAL INSTRUCTIONS: You MUST compare THREE different years of documents.
DO NOT extract or summarize a single document. You are doing COMPARATIVE ANALYSIS ONLY.
Brooklyn Community Board {cb_number} - YEAR-OVER-YEAR COMPARISON
YOU HAVE THREE DOCUMENTS BELOW:
- Document 1: FY2024 needs statement
- Document 2: FY2025 needs statement
- Document 3: FY2026 needs statement
YOUR JOB: Compare how things CHANGED from 2024 → 2025 → 2026.
[Full prompt includes detailed JSON output structure requirements]
Theme Extraction Prompt
Analyzes comparison data to automatically extract and score themes (0-5 depth scale):
- 0 = Not mentioned
- 1 = Basic mention only
- 2 = Mentioned with examples
- 3 = Policy asks or commitments requested
- 4 = Data/evidence cited
- 5 = Structured recommendations or frameworks proposed
View Full Prompt →
Analyze this 3-year community board needs comparison and extract the KEY THEMES.
INSTRUCTIONS:
1. Identify 4-6 major policy themes that appear across FY2024, FY2025, and FY2026
2. For each theme, assign a "depth score" (0-5) for each fiscal year
3. Provide a SHORT label for each theme (2-4 words max)
IMPORTANT RULES:
- Extract themes from actual narrative text, NOT generic categories
- Score based on depth of discussion in policy_changes and emphasis_shifts
- Look for themes that EVOLVED across years
OUTPUT: Valid JSON with themes array including name, scores, and evolution_note
Technology Stack
Backend
- Python 3.10
- OpenAI GPT-4 API
- PDFPlumber for text extraction
- SQLite database
- Requests library
Frontend
- Vanilla JavaScript
- D3.js for visualizations
- Chart.js for charts
- D3-Sankey for flow diagrams
- Custom CSS with playful design
Data Scope
- Borough: Brooklyn (18 Community Boards)
- Fiscal Years: 2024, 2025, 2026
- CBs with complete 3-year data: CB1, CB2, CB3, CB4, CB5 (5 boards)
- Total PDFs analyzed: 52 documents
- Themes extracted: 4-6 per community board, dynamically identified
Future Development
Planned expansions include extending coverage to all NYC boroughs, adding real-time PDF monitoring, implementing topic modeling for cross-board comparison, and building public API access for researchers and journalists.
Needs Assessments are retrieved from NYCPlanning/labs-cd-needs-statements .
The NYC Community Boards Dashboard is an interactive civic tech tool that visualizes trends in Community Board representation, priorities, and governance across New York City. It uses demographic reports, District Needs Statements, and policy themes to provide an accessible overview of how community boards reflect and respond to their neighborhoods. The dashboard empowers residents, advocates, researchers, and policymakers to explore patterns over time and gain insight into equity, engagement, and local decision-making. Get curious about local democracy!
This project is heavily enabled by AI. The website and analysis pipeline were generated by Claude Code and ChatGPT/Codex (vibecoding is so powerful!!). The initial results have very verified and validated, and the work is ongoing. The illustrations of the pigeons are generated by DALL·E 3.
This is an open-source project under the MIT License on GitHub:
github.com/txinjiez/nyc-community-boards-dashboard .
Contributions and feedback are warmly welcomed—feel free to reach out to Tina at tina.zeng@yale.edu.
Tina Zeng is a student at Yale studying Global Affairs + Computing, Culture & Society at Yale. With her passions in civic and public interest technology, democratic innovation and social entrepreneurship, she hopes to leverage technologies like AI for social impact. Tina’s experience serving on Brooklyn Community Board 11 (2023–2024) strengthened her commitment to participatory governance and motivated her to explore ways to make civic institutions more engaging, responsive and equitable.
This research project and dashboard originated as part of her class project in AI for Social Science Methods, taught by Daniel Karrell. Through the Dahl Research Fellowship at Yale’s Institute for Social and Policy Studies, she continues to research Community Boards and expand this dashboard.
She is anticipating submitting a proposal to present at the 2026 School of Data, organized by BetaNYC, where she’s an Associate Board Member (profile). The presentation will be a demo of the dashboard and an explanation of the methodology and development process.
Feel free to reach out to tina.zeng@yale.edu if you’re interested in chatting!