Blackstone (https://github.com/ICLRandD/Blackstone) is an experimental spaCy pipeline and model for performing Natural Language Processing on unstructured legal text. Created by Daniel Hoadley, formerly of the R&D arm of the verbosely named Incorporated Council of Law Reporting for England and Wales.
Blackstone is an adaptation of the general-purpose NLP tool - spacCy - to the particular context of legal texts, which means its model is trained on legal language and designed to process legal language, as opposed to other types of text. The model's been trained on UK court decisions, and seems to work reasonably well on US cases too.
Blackstone aims to help with two tasks that are common to working with legal texts: (1) identifying and classifying entities in court opinions (such as court names, judges, citations, case names) and (2) identifying and classifying common parts of court opinions (such as conclusions, axioms/principles, legal issues and legal tests).
It has good documentation and is easy to install and play around with. Daniel also went to great lengths to share the "why" of Blackstone, which is refreshing to see.
It's unclear what lies ahead for Blackstone (and ICLRandD, for that matter), after Daniel's departure. But it's a great contribution nonetheless. Well worth further exploration and experimentation.
Blackstone is an adaptation of the general-purpose NLP tool - spacCy - to the particular context of legal texts, which means its model is trained on legal language and designed to process legal language, as opposed to other types of text. The model's been trained on UK court decisions, and seems to work reasonably well on US cases too.
Blackstone aims to help with two tasks that are common to working with legal texts: (1) identifying and classifying entities in court opinions (such as court names, judges, citations, case names) and (2) identifying and classifying common parts of court opinions (such as conclusions, axioms/principles, legal issues and legal tests).
It has good documentation and is easy to install and play around with. Daniel also went to great lengths to share the "why" of Blackstone, which is refreshing to see.
It's unclear what lies ahead for Blackstone (and ICLRandD, for that matter), after Daniel's departure. But it's a great contribution nonetheless. Well worth further exploration and experimentation.
Adam