System Requirements Document for simple-fully

1. Introduction

The "AI-Powered Criminal Network Analysis System" is a web application developed for the Smart India Hackathon 2026 by TEAM UPLIFT. Its primary purpose is to integrate fragmented police data into a comprehensive knowledge graph, enabling the extraction of entities and relationships, identification of key persons, and detection of suspicious patterns. The system presents this information through a visual investigator console, facilitating law enforcement and intelligence analysis.

2. System Overview

The system is a fully functional web application that fuses various data sources into a single knowledge graph. It supports entity extraction, relationship mapping, and pattern detection, providing a visual interface for investigators. The application is built using a specified technology stack and adheres to a luxury instrument aesthetic inspired by MARQ by Garmin. It includes features such as data ingestion, graph visualization, key person identification, pattern detection, predictive analysis, and collaborative tools.

2a. Product Interpretation and Delivery Boundary

The application is delivered as a web-based platform with a dark intelligence console design. It supports real-time data processing and visualization, with a focus on precision and control. The system is intended for use by law enforcement agencies and intelligence analysts, providing them with a reliable tool for criminal network analysis. The application includes a comprehensive setup guide and demo mode for ease of use.

2b. Source Content Inventory

Not applicable as no explicit content_source directive is provided.

2c. Page Content and Component Coverage

Dashboard

• Information/State: Overview of system status, recent alerts, and key metrics.
• Primary Actions: Navigate to other sections, view notifications.
• Components: Top bar with shield emblem, title, notification bell, avatar; Sidebar with navigation links.

Cases

• Information/State: List of cases with details and status.
• Primary Actions: View case details, update status, assign roles.
• Components: Case table, filters, action buttons.

Subjects (X/Y)

• Information/State: Profile information for Subject X and Subject Y.
• Primary Actions: Compare subjects, view detailed profiles.
• Components: Profile cards, comparison view.

Evidence Locker

• Information/State: Repository of evidence items with labels.
• Primary Actions: Tag evidence, view details, export items.
• Components: Evidence table, tagging options, export button.

Link Analysis

• Information/State: Interactive graph of entities and relationships.
• Primary Actions: Expand nodes, apply filters, export graph.
• Components: Graph canvas, control panel, export options.

Timeline

• Information/State: Event timeline with date-range slider.
• Primary Actions: Play animation, adjust date range.
• Components: Timeline slider, play button, event markers.

Geospatial Map

• Information/State: Map with incident markers and heatmap.
• Primary Actions: Toggle heatmap, draw radius query.
• Components: Map canvas, control panel, query tools.

Reports

• Information/State: Generate and download case dossiers.
• Primary Actions: Create report, export to PDF.
• Components: Report generation form, export button.

Settings

• Information/State: User preferences and system settings.
• Primary Actions: Update settings, manage roles.
• Components: Settings form, role management tools.

3. Functional Requirements

1. As an officer, I should be able to upload various data formats (CSV/JSON/PDF/TXT/images) for ingestion into the system.

   • Provenance: Explicit
   • Lifecycle Facts: Ingest → OCR/NLP → Entity Extract → Relation Extract → Entity Resolution → Graph
   • Observable Acceptance: Data appears in the review table for further action.

2. As an investigator, I should be able to interact with an interactive graph to explore relationships between entities.

   • Provenance: Explicit
   • Lifecycle Facts: Pan, zoom, click-to-expand, apply filters, export graph
   • Observable Acceptance: Graph updates and exports correctly.

3. As an analyst, I should be able to identify key persons using various centrality metrics.

   • Provenance: Explicit
   • Lifecycle Facts: Metrics calculation, sortable table, badge application
   • Observable Acceptance: Key persons are highlighted based on metrics.

4. As an officer, I should be able to detect patterns and receive alerts for anomalies.

   • Provenance: Explicit
   • Lifecycle Facts: Community detection, anomaly rules, alert generation
   • Observable Acceptance: Alerts are generated with explanations.

5. As an investigator, I should be able to predict links using GNN and confirm or dismiss predictions.

   • Provenance: Explicit
   • Lifecycle Facts: Link prediction, confirmation/dismissal
   • Observable Acceptance: Predicted links are processed correctly.

6. As an officer, I should be able to compare two subjects and export the comparison to PDF.

   • Provenance: Explicit
   • Lifecycle Facts: Profile comparison, live charts, export functionality
   • Observable Acceptance: Comparison view is accurate and exportable.

7. As an analyst, I should be able to view an event timeline and animate graph changes over time.

   • Provenance: Explicit
   • Lifecycle Facts: Timeline interaction, graph animation
   • Observable Acceptance: Timeline and animations function as expected.

8. As an officer, I should be able to locate incidents on a map and perform time-synced queries.

   • Provenance: Explicit
   • Lifecycle Facts: Map interaction, query execution
   • Observable Acceptance: Map updates and queries return correct results.

9. As a user, I should be able to collaborate with others using role-based permissions and live updates.

   • Provenance: Explicit
   • Lifecycle Facts: JWT login, role management, WebSocket updates
   • Observable Acceptance: Collaboration features work with correct permissions.

10. As an investigator, I should be able to understand AI outputs through an explanation panel.

    • Provenance: Explicit
    • Lifecycle Facts: Explanation panel access, feature weight display
    • Observable Acceptance: Explanations are clear and informative.

11. As a user, I should be able to generate and download comprehensive reports.

    • Provenance: Explicit
    • Lifecycle Facts: Report generation, PDF export
    • Observable Acceptance: Reports are generated and downloadable.

4. User Personas

• Admin: Manages system settings and user roles.
• Senior Investigator: Oversees investigations and validates key findings.
• Investigator: Conducts detailed analysis and confirms predictions.
• Analyst: Processes data and identifies patterns.
• Auditor: Reviews system logs and verifies data integrity.

5. Core User Flows

1. Data Ingestion and Review

   • Initiator: Officer
   • Steps: Upload data → System processes data → Officer reviews extractions → Accept/Edit/Reject
   • Outcome: Data is integrated into the knowledge graph.

2. Graph Exploration

   • Initiator: Investigator
   • Steps: Access graph → Interact with nodes → Apply filters → Export graph
   • Outcome: Investigator gains insights into entity relationships.

3. Key Person Identification

   • Initiator: Analyst
   • Steps: Access key persons table → Sort by metrics → Identify key persons
   • Outcome: Key persons are highlighted for further investigation.

4. Pattern Detection and Alerts

   • Initiator: Officer
   • Steps: System detects patterns → Officer reviews alerts → Take action
   • Outcome: Anomalies are addressed based on alerts.

5. Link Prediction and Confirmation

   • Initiator: Investigator
   • Steps: System predicts links → Investigator reviews predictions → Confirm/Dismiss
   • Outcome: Predictions are validated and integrated.

6. Subject Comparison

   • Initiator: Officer
   • Steps: Select subjects → View comparison → Export to PDF
   • Outcome: Comprehensive comparison report is generated.

7. Timeline Animation

   • Initiator: Analyst
   • Steps: Access timeline → Adjust date range → Play animation
   • Outcome: Graph changes are visualized over time.

8. Geospatial Analysis

   • Initiator: Officer
   • Steps: Access map → Perform queries → View results
   • Outcome: Spatial insights are gained from map analysis.

9. Collaboration and Role Management

   • Initiator: Admin
   • Steps: Manage roles → Facilitate collaboration → Monitor updates
   • Outcome: Effective collaboration with role-based permissions.

10. AI Output Explanation

    • Initiator: Investigator
    • Steps: Access explanation panel → Review details
    • Outcome: Investigator understands AI decisions.

11. Report Generation

    • Initiator: User
    • Steps: Generate report → Download PDF
    • Outcome: Detailed report is available for distribution.

6. Visuals Colors and Theme

• Muse: MARQ by Garmin
• Palette:
  • Background: #0E1A2B
  • Surface: #132840
  • Text: #E6EDF5
  • Primary: #4C9BE8
  • Accent: #E8833A
  • Muted: #22496F
• Typography:
  • Headings: Barlow Condensed, uppercase, medium weight
  • Body: Saira
  • Scale: 1.25 modular (48/32/24/16/12)
• Shape Language: Circular gauges, precise data rows, edge-to-edge photography
• Layout: Dashboard-like with ruled panels, dark UI
• Motion: Needle sweeps for data updates, restrained tempo

7. Signature Design Concept

The public entry features a split-screen hero with Subject X vs Subject Y profiles beside a dynamic comparison view. The design emphasizes precision and control, with circular gauges for key metrics and edge-to-edge photography for profile highlights. The dark background is accented with orange and blue, aligning with the luxury instrument aesthetic.

8. Interaction Model & Motion Direction

• Interaction Model: Static (direction)
• Motion Tempo: Restrained
• Hero Dimensionality: Layered_2d
• Landing Hero Motion Brief:
  • Focal Subject: Subject X vs Subject Y profiles
  • Thesis: Input (select subjects) → Transformation (dynamic comparison) → Outcome (visual insights)
  • Motion Vocabulary: Needle sweeps, slow transitions
  • First Frame: Split-screen with highlighted profiles
  • Reduced-Motion State: Static comparison view

9. Non-Functional Requirements

• Performance: The system must handle large datasets efficiently, with real-time updates and low latency.
• Security: Implement JWT authentication and role-based access control to ensure data protection.
• Scalability: The application should support scaling to accommodate increased data and user load.
• Reliability: Ensure high availability and fault tolerance with in-memory fallback and Docker orchestration.

10. Tech Stack

• Frontend: React, TypeScript, Vite, Tailwind, Cytoscape.js, Recharts, Leaflet, OpenStreetMap
• Backend: FastAPI, NetworkX, PyTorch Geometric, scikit-learn, spaCy
• Database: Neo4j, PostgreSQL
• Deployment: Docker Compose

11. Assumptions and Constraints

• Assumptions: Users have basic technical proficiency and access to necessary hardware.
• Constraints: The application must adhere to the specified technology stack and design guidelines.

12. Glossary

• GNN: Graph Neural Network
• JWT: JSON Web Token
• OCR: Optical Character Recognition
• NLP: Natural Language Processing
• AI: Artificial Intelligence