Cara Bilang - Colloquial English Search Engine for Indonesians
·
2 min read
Search-first platform helping Indonesians find natural English expressions for real situations. Features intent-based search powered by Meilisearch and high-quality conversational content.
Overview
Cara Bilang (Indonesian for “How to Say”) is a search-first platform designed to bridge the gap between formal English proficiency and real-world conversational fluency for Indonesian speakers.
Traditional language tools often focus on grammar rules or vocabulary lists, leaving learners struggling with situatinal context. Cara Bilang reverses this by starting with the user’s intent—allowing them to describe what they want to say in natural Indonesian and receiving idiomatic, contextually appropriate English expressions in return.
Key Features
Intent-Based Search Engine
- Natural Language Lookup: Search for expressions using Indonesian or English situational descriptions.
- High Recall: Powered by Meilisearch with custom indexing strategies to capture colloquial nuances.
- Semantic Mapping: Designed to handle ambiguous intents and provide the most culturally relevant matches.
Rich Conversational Data
- Phrase Patterns: Reusable linguistic structures that can be adapted to various contexts.
- Authentic Examples: Multi-line conversation snippets showing how phrases are used in real dialogue.
- Social Discourse: Nuanced content covering everything from professional refusal to casual social reactions.
Robust Engineering Architecture
- Clean Architecture: A maintainable Go backend following domain-driven design principles.
- Dual-Store Strategy: Uses PostgreSQL as the canonical source of truth and Meilisearch for ultra-fast, search-optimized delivery.
- Content Operations: A dedicated Admin UI for managing complex JSON-based content batches and publishing workflows.
Tech Stack
- Backend: Golang (Gin framework) with
pgxand Clean Architecture - Frontend: React 19 with TanStack (Router, Query) and Vite
- Database: PostgreSQL (Canonical storage)
- Search Engine: Meilisearch
- Styling: Tailwind CSS
- Infrastructure: Docker-compose for local development, Python-powered embedding service for semantic experiments
Why This Project?
Most Indonesians learn “textbook English” but struggle with the “how do I say this naturally?” problem. Cara Bilang addresses this by:
- Prioritizing Intent: Focusing on what the user wants to achieve (e.g., “how to politely say no to a boss”) rather than just translating words.
- Designing for Speed: Providing a search-first interface that feels like a utility tool rather than a slow course.
- Operational Excellence: Building a system that can handle thousands of hand-curated intents with a structured ingestion pipeline.