Read on Twitter
As someone that spends quite a lot of time thinking about NLP in the context of legal information, I often find myself feeling slightly jealous (for want of a better word) of those doing NLP in the scientific and medical domains.
The reason for this is that, in contrast with the legal domain, the scientific community has always been pretty good at building and maintaining lots and lots of high quality lists and ontologies, for example the UMLS https://www.nlm.nih.gov/research/umls/index.html and MeSH https://www.nlm.nih.gov/mesh/meshhome.html
One of the many purposes NLP aims to serve is the discovery of ways to connect language to knowledge in order to help us understand the information we have and to point us towards other sources of information that are relevant to the things we care about. Good ontologies are key.
In the legal domain (and certainly in E&W) we lack these ontologies. It’s true that the legal publishers maintain their own ontologies, but those ontologies suffer from two major flaws. The first is that they aren’t openly accessible.
They can only be consumed through the prism of the product they were built to support. The second flaw is that even if those ontologies were openly available, the ontology from one publisher is unlikely to align with the others, even though they purport to model the same “things”
The purpose of an ontology is that it should ideally provide an authoritative portal to a single source of truth that maps variable descriptions of the same thing or concept to their canonical and normalised root.
So here’s my idea. The legal publishers in EW should join together, in partnership with government and perhaps others in the private sector and agree to build and maintain two connected super ontologies.
The first would be the Case Law Ontology. This would catalogue all cases decided in the senior courts starting from cases reported in the Nominates through to the present day, capturing common variations of the name; known citations; dates; court; high level subject matter;
... key legal concepts engaged; and, crucially, where that case is available to be accessed. The second would be the Legal Concept Ontology. This would effectively act as an agreed dictionary of legal concepts that would map concepts and variations thereof.
All very blue sky and probably unrealistic (though far from impossible) for a whole raft of reasons. But, I think it would be incredible if it could be pulled off.
