Maichi Parser: Cleaning up the "nightmare" of address data in Vietnam
Anyone involved in e-commerce, logistics, or data management has likely faced at least once the "nightmare" of poorly formatted addresses from customers. Entering data like this: "58 p. lê văn hiến đ.thắng btl hnoi" would have even the postal system "screaming," let alone the operations staff.
- Using digital maps (Geocoding)? If a customer misspells or is from a remote area, the system just "clogs up."
- Using AI/LLM? The risk of "hallucination" is extremely high; occasionally, a playful AI might fabricate a non-existent neighborhood to please you.
That's why Maichi Parser was created – a machine specifically designed to "translate" and standardize the informal address writing habits of Vietnamese people, turning junk data into gold.
What "weapons" does Maichi Parser possess?
Simply put, you throw a jumbled string of characters at Maichi, and the system will automatically refine and clearly parse it: identifying the Province, District, Commune; and extracting the street name and house number in a very tidy manner.
The key features that differentiate Maichi:
- Smart & Comprehensive Standardization: It automatically understands that "hnoi" means Hanoi, and "btl" means Bac Tu Liem. If a customer omits administrative levels, Maichi deduces and fills in the gaps. Notably, the system automatically assigns the standard Administrative Code of the General Statistics Office (GSO Code) for each ward/commune. Anyone who prepares government reports or synchronizes delivery data will find this feature "life-saving."
- 100% Consistent (No guesswork from AI): It operates on a Deterministic Engine (hard rules) rather than LLM, combined with an enormous data dictionary. The output from Maichi is accurate and absolutely consistent. What you see is what you get; there will never be random address guessing to trick the system.
- Huge throughput & Super low latency: Capable of processing up to 12,000 addresses/second with a P99 latency of under 10ms. No worries about congestion, no matter how large the integration system is.
Easy Integration - Usable by Everyone!
For technical folks (Developers):
- Providing ready-to-use Cloud API (REST/JSON and high-performance gRPC), just one call and you get the data.
- Need absolute security for your business? There's an On-Premise version (K8s friendly) that can be installed directly on an internal server. The system keeps running smoothly even if the undersea cable is cut.
For Shop owners, Operations staff (no coding knowledge required):
- We have built a file cleaning tool ready for you. Just upload your messy Excel/CSV file containing thousands of orders to the website, and the system will automatically clean it up and provide you with a new, pristine file. It's a breeze!
Illustrating a Real Case
Your input: "58 p. lê văn hiến đ.thắng btl hnoi"
The output returned by Maichi (data standardized for software):
- Province/City: Hanoi (GSO Code: 01)
- District: Bac Tu Liem (GSO Code: 021)
- Ward/Commune: Duc Thang (GSO Code: 00602)
- Street Name/House Number: 58 P. Le Van Hien
Processing 1-2 orders can be done visually, but for thousands of orders, let the machines handle it.
Ready to optimize your system?
Reviews & comments