Legislate’s knowledge graph for document generation United States patent 11,087,219 was issued on the 10th of August 2021. Prior to Legislate’s innovation, document automation was limited to simple parts of document generation or the process surrounding the generation of the documents such as e-signatures. This article explains the key takeaways from the patent, its impact and what it means for the future of document automation.
What is document automation?
Document automation is a field which focuses on automating the steps required to generate a document. Document automation can be applied to any type of document and has been applied to generating invoices, reports and more recently tweets and blog posts thanks to machine learning.
Why focus on document automation?
Whilst document automation is widely adopted, it tends to focus on simple use cases. For example, filling in information within a static structure or automating the process around the document. For example, the process of printing, signing and scanning is no longer necessary with electronic signatures. Legislate’s mission is to automate more challenging and dynamic document generation tasks such as the creation and negotiation of contracts.
Why is smart document automation so difficult?
Documents contain information which is unstructured and difficult for algorithms to understand. This means that it is harder for algorithms to find consistent patterns in text because meaning can be expressed in many equivalent versions. Structuring data in documents using approaches like legal schema facilitate the automation of processes around documents easier but don’t address the challenges with generating documents. Legal documents are challenging in their own right due to legalese and the nature of legal data. One solution for addressing the unstructured nature of documents is to represent them as knowledge graphs which is the basis of Legislate’s patent.
What is a knowledge graph?
A knowledge graph is a type of database which is highly suited to unstructured yet connected information because it stores data as relations as opposed to in tables. Moreover, knowledge graphs can encode expertise via an ontology or rules to map relations and deduce meaning from patterns.
How is a document modelled in a knowledge graph?
In order for algorithms to understand a document’s information, the information needs to be machine readable. One way of helping algorithms understand what a document contains is to highlight the key elements of its structure and how they are connected, for example in a knowledge graph.
Structuring the information of a document means it can be intelligently managed and searched. When a knowledge graph is taught concepts via rules it can derive new information or organise it in more relevant ways. Google is able to recognise certain types of information (e.g. a FAQ or How To) and display it in a more relevant way (e.g. knowledge panel) because the information on the web pages is structured into Google’s own knowledge graph.
How is Legislate using knowledge graphs?
Legislate’s patent protects a way of structuring the information of a document in a knowledge graph so that complex documents can be negotiated by multiple parties and generated automatically.
Moreover, since Legislate’s knowledge graph models domain expertise, it can validate the structure of documents and prevent non-coherent values from being inserted. The same approach can be applied to reconcile documents provided by different parties trying to negotiate terms.
Legislate’s knowledge graph can also use domain expertise, modelled as rules, to derive concepts and present the document’s information at different levels of abstraction. This means that a person can quickly grasp the key takeaways of the document without needing to read it granularly. As a result, it can answer questions such as “what are my obligations” or “am I allowed pets in my rental property?”.
What does this mean for the future of document automation?
Smart documents have until now been a challenge because of the absence of a common framework to structure a document’s key concepts. Initiatives like legal schema enable certain types of document information to be structured post-generation. Legislate enables smart documents to be generated and updated without compromising the consistency of its structure thanks to a knowledge graph. This means that document automation can move beyond simple, static documents to complex and dynamic ones.
What does this mean for the unlawyered?
Complex documents have typically relied on human expertise to be generated. Contracts in particular have relied on lawyers to be created and tailored. Thanks to Legislate, anyone can configure robust and safe contracts by themselves and understand what they are doing. Moreover, contracts are then stored in their Legislate vaults which means they can be accessed on the fly by contract parties and intelligently managed and searched. Finally, Legislate’s knowledge graph provides instant visibility and insights into contracts to help businesses make better decisions.
Legislate is an early stage legal technology startup which allows large landlords, letting agents and small businesses to easily create, sign and manage contracts that are prudent and fair. Legislate’s platform is built on its patented knowledge graph which streamlines the contracting process and aggregates contract statistics to quickly unlock valuable insights. Legislate’s team marries technical and legal expertise to create a painless, smart contracting experience for its users. Legislate is backed by Parkwalk Advisors, Perivoli Innovations and angel investors.