ryanctlin / EU_Legislation_NLP

Natural language processing of EU legislation references

EU_Legislation_NLP

Natural language processing of EU legislation references using IBM's Watson Cloud platform

The program accepts EU legislation as an EUR-Lex URL (e.g. "https://eur-lex.europa.eu/legal-content/EN/TXT/?qid=1536601415699&uri=CELEX:32018R0643") and returns a list of all Regulations and Directives referenced in the legislation text, decomposed into relevant components: Legislation Type, Number, Year, Article, Paragraph, Point.

The project consists of three phases:

1. Accept URL input and extract legislation text using BeautifulSoup
2. Identify and isolate whole references (e.g. "point 25 of Article 2 of Directive 2009/73/EC") from legislation by training a custom Natural Language Understanding (NLU) model for Watson Cloud
3. Decompose references into components (e.g. "POINT": "point 25", "ARTICLE": "Article 2) by training a second custom Natural Language Understanding (NLU) model for Watson Cloud