[LEGAL AI] Part 2-1:Gathering Essential Legal Data

-Daniel

Hey everyone!

I’ve decided to explore the world of Legal AI, and it’s been quite an adventure! 🚀 Living in Korea, I figured starting with our own laws made sense. Plus, who doesn’t enjoy a good challenge, right?

The Grand Quest for Legal Data

First, I needed to identify what legal data to feed into my AI. Not all laws are created equal (who knew?). Here’s a breakdown of the different types of legal documents I considered:

• Current Laws (현행법령): These are the laws currently in effect and are crucial for ensuring our AI provides the most current answers. Outdated laws won’t suffice!
• Recently Amended Laws (최신법령): These include laws that have been updated or introduced recently. The legal landscape changes quickly, so these are nearly as important as current laws.
• Historical Laws (연혁법령): These are previously repealed laws that were once in effect. They’re useful for understanding how laws have evolved over time.
• Modern Legal Codes (근대법령): Old laws not relevant today. While they aren’t directly useful, they can be interesting for those who love histor.y
• International Treaties (조약): Agreements between countries. These are essential when dealing with cases involving foreign elements.

Prioritizing Like a Pro

After much thought (and coffee), I established a priority list:

• Top Priority: Current Laws and Recently Amended Laws
• Middle Ground: Historical Laws and International Treaties
• Low Priority: Modern Legal Codes (unless you’re a history enthusiast!)

Why this order? The AI needs to provide accurate, relevant answers based on today’s laws. However, historical context can be beneficial too.

Data Sources: API vs. Web Crawling

Here’s where things got interesting—and by interesting, I mean mildly frustrating.

• Option 1: Official APIs
  I discovered that data.go.kr offers APIs related to legal information. Perfect! Except, you need to apply for access and wait about a week. Being impatient (and starting this project over a holiday), waiting wasn’t an option.
  Korean: https://open.law.go.kr/LSO/openApi/guideList.do# / https://www.data.go.kr/iim/api/selectAPIAcountView.do
  English: https://openlaw.klri.re.kr/service/intro/introinfo.do
• Option 2: Web Crawling
  “How hard can it be?” I thought. Spoiler alert: harder than you’d expect. The National Law Information Center’s website (laws.go.kr) features numerous pages with slightly different structures. Additionally, the site was slow to load and frequently disconnected. But, as they say, nothing worth doing is easy, right?
  https://www.law.go.kr/

After wrestling with HTML structures and handling constant timeouts, I managed to scrape the data I needed. Victory!

Lessons Learned (So Far)

• Patience is a virtue: Waiting for API access might have saved me some headaches.
• Flexibility is key: Sometimes you have to pivot and find alternative solutions.
• The journey is half the fun: Despite challenges, overcoming them was satisfying.

What’s Next?

I’m planning to fine-tune my AI with the data I’ve gathered to see how well it can answer legal questions. Eventually, I hope to expand this project to include laws from the U.S. and other countries. Global Legal AI domination? Perhaps. 😉

Why Am I Sharing This?

I wanted to share this journey with all of you. If you’re interested in AI, law, or complex projects, let’s connect! Who knows? Maybe we’ll collaborate on something amazing.

Feel free to comment or reach out if you have thoughts or tips. And if you’re a recruiter who stumbled upon this post—well, you know where to find me! 😉

Cheers!