Epic 1 - Knowledge Graph

LWKG Knowledge Graph Platform

Comprehensive health and wellness knowledge base with semantic search and intelligent recommendations

Test Coverage

90%

20+ Unit Tests

Foods Database

5,120+

Nutrition Data

Recipes

2,500+

Pre-Made Plans

Activities

100+

Exercise Types

Core Capabilities

Food Database - 5,120+ foods with complete nutritional information (calories, macros, micronutrients)
Recipe Library - 2,500+ pre-made recipes with ingredients, instructions, and nutrition calculations
Activity Registry - 100+ activities with MET values for calorie burn estimation
Semantic Search - pgvector-powered semantic embeddings for intelligent food/recipe discovery
Nutritional Analysis - Automatic nutrition calculation based on recipe ingredients
Allergen Tracking - Complete allergen information for all foods and recipes
Cultural Preferences - Cuisine and cultural tagging for personalized recommendations
Ingredient Composition - Detailed ingredient mapping with quantities and preparation notes

Food Database Features

Comprehensive Nutrition Data

Each food item includes:

Calories per 100g
Protein, carbs, fat, fiber
Micronutrients (vitamins, minerals)
Glycemic index/load
Allergen information

Food Classification

Organized by:

Food groups
Diet types (vegan, keto, etc.)
Cultural cuisines
Processing level
Health considerations

Semantic Embeddings

pgvector integration:

1536-dimensional embeddings
Similarity-based search
Fast retrieval (<100ms)
Nearest neighbor queries
Custom similarity scoring

Data Freshness

Maintained by:

Regular nutrition database sync
Version tracking
Change history audits
Expert review workflows
User feedback integration

Recipe System

                Recipe Management: Complete recipe management system with automated nutrition calculation from ingredient composition. Each recipe includes preparation time, difficulty level, cuisine type, and serving information. Recipes are automatically tagged with allergens based on ingredients.
            

Recipe Features:

Ingredient Tracking - Precise ingredient quantities with unit conversion support
Nutrition Calculation - Automatic nutrition calculation based on ingredients
Difficulty Levels - Easy, Medium, Hard, Professional classifications
Preparation Time - Cooking and prep time estimates
Yield Tracking - Number of servings and portion sizes
Instructions - Step-by-step cooking instructions
Cuisine Types - Indian, Italian, Mediterranean, Asian, etc.

Activity & Exercise Database

Activity Metrics

Each activity includes:

MET value for calorie burn
Intensity level
Duration range
Equipment needed
Difficulty classification

Activity Categories

Aerobic exercises
Strength training
Flexibility & stretching
Sports & recreation
Occupational activities

Search Capabilities

                SEMANTIC Vector similarity search with configurable thresholds

                KEYWORD Full-text search across names, descriptions, and tags

                FILTER Advanced filtering by nutrition, allergens, diet type, and more

                COMBINATION Hybrid search combining semantic and keyword approaches

API Endpoints

                GET /api/v1/foods?q=chicken&limit=10 - Search foods

                GET /api/v1/foods/{food_id} - Get food details

                GET /api/v1/recipes?cuisine=indian&limit=20 - Search recipes

                GET /api/v1/recipes/{recipe_id} - Get recipe details

                GET /api/v1/activities?mets=5-7 - Search activities by intensity

                GET /api/v1/activities/{activity_id} - Get activity details

                POST /api/v1/foods/semantic-search - Semantic search endpoint

                GET /api/v1/foods/allergens/{allergen} - Find foods with specific allergen

Performance Metrics

Semantic Search Latency - <100ms for 5,120 foods using pgvector
Keyword Search - <50ms for full-text search
Recipe Lookup - <10ms direct ID lookup
Concurrent Users - 1,000+ simultaneous searches supported
Data Consistency - Eventual consistency with <1 second propagation