Automating Legal Information and Knowledge Extraction (ALIKE)

The ALIKE project investigates how AI can generate valid statements of law by extracting and processing information from authoritative legal sources. Focusing on the complex yet circumscribed domain of EU law, the project aims to produce a prototype legal information system capable of generating legal textbooks.

Three research phases

The project is structured into three interrelated phases to bridge the gap between raw data and valid legal knowledge:

Assemble

Uses network analysis and Natural Language Processing (NLP) to extract information from databases of EU legislation and case law. It models "triads" consisting of relationships between an “informant” (citing source), a rule interpretation (case law text that elaborates on legal meaning), and an “authority” (cited source).

Represent

Leverages state-of-the-art Large Language Models (LLMs) to represent these assembled rules in readable natural language. To ensure transparency, every generated text must be anchored to identifiable legal sources.

Validate

Employs legal experts (professionals and researchers) to test the AI's output. Experts will attempt to distinguish AI-generated text from human-authored textbook passages and assess the legal correctness of the machine output.

ASSEMBLE: Uses network analysis and Natural Language Processing (NLP) to extract information from databases of EU legislation and case law. It models "triads" consisting of relationships between an “informant” (citing source), a rule interpretation (case law text that elaborates on legal meaning), and an “authority” (cited source).
REPRESENT: Leverages state-of-the-art Large Language Models (LLMs) to represent these assembled rules in readable natural language. To ensure transparency, every generated text must be anchored to identifiable legal sources.
VALIDATE: Employs legal experts (professionals and researchers) to test the AI's output. Experts will attempt to distinguish AI-generated text from human-authored textbook passages and assess the legal correctness of the machine output.

ASSEMBLE: Uses network analysis and Natural Language Processing (NLP) to extract information from databases of EU legislation and case law. It models "triads" consisting of relationships between an “informant” (citing source), a rule interpretation (case law text that elaborates on legal meaning), and an “authority” (cited source).
REPRESENT: Leverages state-of-the-art Large Language Models (LLMs) to represent these assembled rules in readable natural language. To ensure transparency, every generated text must be anchored to identifiable legal sources.
VALIDATE: Employs legal experts (professionals and researchers) to test the AI's output. Experts will attempt to distinguish AI-generated text from human-authored textbook passages and assess the legal correctness of the machine output.

ASSEMBLE: Uses network analysis and Natural Language Processing (NLP) to extract information from databases of EU legislation and case law. It models "triads" consisting of relationships between an “informant” (citing source), a rule interpretation (case law text that elaborates on legal meaning), and an “authority” (cited source).
REPRESENT: Leverages state-of-the-art Large Language Models (LLMs) to represent these assembled rules in readable natural language. To ensure transparency, every generated text must be anchored to identifiable legal sources.
VALIDATE: Employs legal experts (professionals and researchers) to test the AI's output. Experts will attempt to distinguish AI-generated text from human-authored textbook passages and assess the legal correctness of the machine output. 

External