System Requirements Document (SRD)

Project Name: fiery-rumor

1. Introduction

The fiery-rumor project is an AI-driven crisis rumor verification engine designed to help users verify the authenticity of rumors during critical events. By leveraging advanced AI models and APIs, fiery-rumor processes user-submitted rumors, retrieves relevant news articles, and provides a verdict on the rumor's authenticity. The system is designed to be user-friendly, transparent, and efficient, catering to the needs of users in India (IN) and beyond.

This document outlines the system requirements for fiery-rumor, including functional and non-functional requirements, user personas, design concepts, and technical specifications.

2. System Overview

The fiery-rumor system operates through a 3-stage pipeline:

1. Stage 1: Translation and Entity Extraction

   • Accepts user input (rumor text).
   • Detects the source language and translates it to English if necessary.
   • Extracts structured information such as event, location, date, and keywords.

2. Stage 2: Live News Retrieval

   • Uses the extracted information to query a live news API (e.g., NewsData.io).
   • Retrieves the top 3 related articles based on the rumor's context.

3. Stage 3: Final Verification

   • Analyzes the rumor and retrieved articles using an AI model.
   • Provides a verdict (TRUE, FAKE, or UNVERIFIED), a confidence score, and a brief summary.

The system ensures transparency by displaying a "Transparency List" of clickable news sources and URLs.

3. Functional Requirements

• As a User, I should be able to paste a rumor into the input box.
• As a User, I should be able to view the system's verdict (TRUE, FAKE, or UNVERIFIED) along with a confidence score.
• As a User, I should be able to see a brief AI-generated summary explaining the verdict.
• As a User, I should be able to view a "Transparency List" of related news articles with clickable titles and URLs.
• As a User, I should be able to see the system handle rumors in multiple languages.
• As a User, I should be able to view a clean and responsive UI on both desktop and mobile devices.
• As a User, I should be able to selectively re-run the verification process for a previously submitted rumor to regenerate the results.

4. User Personas

4.1 General User

• Description: Individuals seeking to verify the authenticity of rumors during crises or critical events.
• Goals:
  • Quickly verify rumors.
  • Access reliable sources of information.
  • Understand the reasoning behind the system's verdict.

4.2 Admin

• Description: System administrators responsible for maintaining the fiery-rumor platform.
• Goals:
  • Monitor system performance.
  • Manage API keys and configurations.
  • Ensure data security and compliance.

5. Visuals Colors and Theme

The fiery-rumor project will use a unique color palette inspired by the urgency and reliability associated with crisis verification.

• Background: #1A1A2E (Deep Midnight Blue)
• Surface: #16213E (Dark Navy Blue)
• Text: #EAEAEA (Soft White)
• Accent: #F05454 (Fiery Red)
• Muted Tones: #6C757D (Muted Gray)

This palette conveys a sense of trust, urgency, and professionalism, aligning with the project's purpose.

6. Signature Design Concept

Interactive Crisis Map Landing Page

The fiery-rumor homepage will feature an interactive crisis map as its centerpiece.

• Design Details:

  • A dynamic world map with glowing red hotspots indicating regions with active rumors.
  • Users can hover over a hotspot to see a tooltip with a brief description of the rumor (e.g., "Flood in Mumbai, India").
  • Clicking on a hotspot will auto-fill the rumor input box with the selected rumor for verification.

• Animations and Transitions:

  • The map will have a subtle pulsing animation for active hotspots.
  • Smooth zoom-in and zoom-out transitions when users interact with the map.

• Micro-Interactions:

  • When hovering over a hotspot, the tooltip will fade in with a slight bounce effect.
  • Clicking on a hotspot will trigger a ripple effect on the map.

This design concept ensures an engaging and memorable first impression while providing users with immediate value.

7. Non-Functional Requirements

• The system must respond to user queries within 5 seconds.
• The system must support at least 10,000 concurrent users.
• The system must comply with data privacy regulations (e.g., GDPR, Indian IT Act).
• The system must be accessible on both desktop and mobile devices.
• The system must store API keys securely in a .env file.
• The system must allow selective re-runs of the verification process for previously submitted rumors.

8. Tech Stack

Frontend

• HTML5, CSS3 (modern, responsive UI)
• Vanilla JavaScript

Backend

• Python 3.11+
• Flask

Database

• SQLite for storing request history, results, and confidence scores

AI Models

• Google Gemini (gemini-3-flash-preview) for translation and entity extraction
• Groq SDK for final verification using Llama 3 70B/8B

APIs

• NewsData.io API for live news retrieval

AI Tools

• LangChain
• LiteLLM for LLM routing

Orchestration

• Docker and docker-compose for local orchestration

9. Assumptions and Constraints

• The system assumes that users will provide rumors in text format.
• The system assumes that API keys for external services (e.g., NewsData.io) will be manually added to the .env file by the admin.
• The system is constrained by the rate limits of external APIs.
• The system assumes a stable internet connection for accessing external APIs.
• The selective re-run capability assumes that previously submitted rumors and their results are stored in the SQLite database.

10. Glossary

• Rumor: A piece of unverified information submitted by the user for verification.
• Transparency List: A list of related news articles retrieved during the verification process.
• Verdict: The system's determination of whether a rumor is TRUE, FAKE, or UNVERIFIED.
• Confidence Score: A numerical representation (0-100%) of the system's confidence in its verdict.
• Hotspot: A region on the interactive map with active rumors.
• Selective Re-Run: A feature allowing users to re-trigger the verification process for a previously submitted rumor.

This concludes the updated System Requirements Document for the fiery-rumor project. Let me know if you need further refinements, Prathamesh!