Dewey Decimal Classification and Legal Taxonomies

In the late 1970s and early 1980s, I was a library prefect in school, filing books on shelves following the Dewey Decimal System. My local public library was perpetually behind in re-shelving.  On the sly, I used to secretly re-shelve books during drop-in visits. When I was 16 I landed a clerk job at an Atlanta law firm.  When the law librarian went out on maternity leave, I stepped in as interim librarian and learned the Lexis-Nexis DOS lookup system, pulled books from local law libraries, and re-shelved the books that stood taller than a sitting associate's head on Monday mornings. 

Dewey decimal classification is a method of assigning relative location based on discipline then topic. Law is classification 340 within the Social Sciences (300). First copyrighted in 1875 by Melvil Dewey after receiving input to his concept from many librarians (despite his notoriously difficult personality), by the 1920s the classification system was ubiquitous in the United States and remains in force today. 

In 2009 Jun Wang, Department of Information Management at Peking University, published in the Journal of the American Society for Information Science and Technology [60(11):2269-2286] a human-assisted algorithm approach incorporating Dewey that allowed 90% accuracy within 3 interactions.  An Internet search shows additional published research on incorporating Dewey into algorithms through 2019.  I am not aware whether any legal technologies leveraging AI are taking advantage of the Dewey Decimal Classification system for the topical assignment portion and would welcome intel on that subject.

In every legal operations position I have been asked to solve a long-standing debate on a document (or contract) management taxonomy. I understand the siren song but know better than to take it on.  Any legal tech company addressing the taxonomy issue needs to come up with standardization.  

The Dewey Decimal Classification has withstood 140 years.  It is the system every living professional was born into and accepts as part of the natural order, like gravity.  The framework is pervasive. One still has to come up with a secondary classification for document type (agreement, brief, memo, etc.) or better yet, a tagging approach to accommodate the fact that many documents contain layers (a master agreement with a digital processing addendum or a personal services contract with an embedded NDA, for example).  However, following the Dewey Decimal System for topic is a good bet for industry adoption.