srs_reciping

Reciping-SRS.md

Software Requirements Specification (SRS)

Document ID: SRS-RECIPING-001
Revision: 1.0
Date: 2025-06-26
Standard: ISO/IEC/IEEE 29148:2018

---

1. Introduction

1.1 Purpose

This document defines the requirements for Reciping, a microservices-based Korean recipe sharing platform with AI-powered recommendations, logging infrastructure, and cloud-native deployment in accordance with the ISO/IEC/IEEE 29148:2018 standard.

1.2 Scope

• Recipe Management: CRUD operations with social features (likes, bookmarks, comments)
• AI-Powered Recommendations: NeuMF model-based personalized suggestions with MLflow tracking
• Recipe Search: Search functionality with DishType category filtering
• Advertisement System: A/B testing framework with logging and monitoring
• Logging Infrastructure: Real-time log collection via Kafka and EFK stack with structured logging
• Cloud Infrastructure: AWS EKS/ECS deployment with Terraform IaC and automated CI/CD pipeline

1.3 Definitions, Acronyms, Abbreviations

Term	Definition
MSA	Microservices Architecture
NeuMF	Neural Matrix Factorization for collaborative filtering
EFK	Elasticsearch, Fluentbit, Kibana logging stack
MLflow	Machine learning lifecycle management platform
IaC	Infrastructure as Code
A/B Test	Comparative experiment between advertisement variants

---

2. Stakeholders

Role	Department	Responsibility
Planning Manager (PM)	Project Management	Requirements gathering and prioritization
Backend Development Lead	Backend Team	API design and microservices architecture
ML Engineer	Data Science Team	NeuMF recommendation system and ML pipeline
DevOps Engineer	Infrastructure Team	Cloud infrastructure, CI/CD, and monitoring
Logging Engineer	Platform Team	Logging infrastructure and data pipeline

---

3. System Context and Interfaces

User → React Web App → Backend Services (EKS)
                    ├── Recipe Service
                    ├── User Service  
                    ├── Advertisement Service (A/B Testing)
                    └── Search Service
                            ↓
                    ML Service (ECS) → NeuMF Model
                            ↓
                    Data Layer (PostgreSQL, MongoDB, S3)
                            ↓
                    Logging Pipeline (Kafka → EFK Stack)

---

4. Specific Requirements

4.1 Functional Requirements

ID	Title	Priority	Acceptance Criteria	Owner
REQ-F-001	Recipe CRUD with Social Features	Must Have	Users can create/view recipes, like/bookmark, add comments with DishType categorization	Backend Lead
REQ-F-002	NeuMF-based Recipe Recommendations	Must Have	ML model provides personalized recommendations with 75%+ accuracy	ML Engineer
REQ-F-003	Recipe Search with DishType Filtering	Must Have	Users can search recipes and filter by DishType categories	Backend Lead
REQ-F-004	Advertisement A/B Testing Framework	Must Have	System serves different ad variants with statistical tracking and performance monitoring	Backend Lead
REQ-F-005	Structured Logging with Business Context	Must Have	All services generate structured logs with business context via Kafka and EFK stack	Logging Engineer
REQ-F-006	Infrastructure as Code Deployment	Must Have	GitHub Actions deploys infrastructure changes via Terraform	DevOps Engineer
REQ-F-007	ML Pipeline Automation	Should Have	Automated data processing, training, and model deployment	ML Engineer

4.2 Non-Functional Requirements

ID	Requirement	Target Value	Owner
REQ-NF-001	API Response Time (p95)	≤ 800ms	DevOps Engineer
REQ-NF-002	System Availability	≥ 99.9% monthly	DevOps Engineer
REQ-NF-003	Log Processing Latency	≤ 3 seconds	Logging Engineer
REQ-NF-004	ML Pipeline Execution Time	≤ 10 minutes	ML Engineer
REQ-NF-005	ML Model Response Time	≤ 200ms	ML Engineer
REQ-NF-006	CI/CD Pipeline Execution	≤ 5 minutes	DevOps Engineer

---

5. Technology Stack

Frontend:
  - React Web Application

Backend Services (EKS):
  - Java Spring Boot (Recipe, User, Advertisement, Search)
  - FastAPI (ML Recommendation Service on ECS)

Data Storage:
  - PostgreSQL 
  - MongoDB 
  - S3 (Media files, model artifacts)

ML Infrastructure:
  - PyTorch (NeuMF model)
  - MLflow (Experiment tracking)
  - Prefect (Pipeline orchestration)

Logging & Monitoring:
  - Kafka (Log streaming)
  - Fluent Bit (Log collection)
  - Elasticsearch (Log storage)
  - Kibana (Log visualization)
  - Prometheus + Grafana (Metrics)

Infrastructure:
  - AWS EKS (Microservices)
  - AWS ECS (ML containers)
  - Terraform (Infrastructure as Code)
  - GitHub Actions (CI/CD)

---

6. Core Data Models

// Recipe Classification
public enum DishType {
    ALL, BANCHAN, MAIN_DISH, SOUP, STEW, DESSERT,
    NOODLE_DUMPLING, RICE_PORRIDGE, FUSION, KIMCHI_SAUCE,
    SEASONING, WESTERN, SALAD, SOUP_WESTERN, BREAD,
    SNACK, BEVERAGE, ETC;
}

// Logging System
public enum LogType {
    CREATE_LIKE, DELETE_LIKE, GET_LIKE, RECIPE_CREATE,
    RECIPE_VIEW, USER_SIGNUP, USER_LOGIN, SEARCH_QUERY,
    AD_CLICK, AD_VIEW;
}


// Advertisement A/B Testing
public enum AdVariant {
    VARIANT_A, VARIANT_B;
}


// System Management
public enum UserRole {
    ADMIN, USER, GUEST;
}

---

7. Traceability Matrix

Requirement ID	Module	Implementation Component	Test Validation
REQ-F-001	Recipe Service	RecipeController	Unit + Integration
REQ-F-002	ML Service	NeuMFRecommendationEngine	Model Accuracy
REQ-F-003	Search Service	RecipeSearchService	Search Relevance
REQ-F-004	Advertisement Service	ABTestManager	Statistical Testing
REQ-F-005	Logging Service	StructuredLoggerService	Log Integrity
REQ-F-006	Infrastructure	TerraformModules	Deployment Testing
REQ-F-007	ML Pipeline	PrefectWorkflow	Pipeline Execution

---

Document Prepared by: Backend Development Lead
Technical Review: ML Engineer, DevOps Engineer, Logging Engineer
Final Approval: Planning Manager (PM)

---

This SRS document defines the core requirements for Reciping platform, focusing on implemented features including A/B testing framework, logging infrastructure, and cloud-native ML operations.