Contract automation through machine learning reduces manual tasks associated with contract generation and management.
How natural language processing is transforming contract management
SHARE THIS ARTICLE
Natural language processing (NLP) has become integral to contract management and many other legal and business processes in recent years. NLP is essential to numerous features professionals take for granted in contract lifecycle management (CLM) software. While the technology has a long way to go before it mimics individuals perfectly, commercial and retail applications of NLP are proliferating. Valuate predicts the NLP market size will reach $27.6 Billion by 2026 from $9.9 Billion in 2020.
What is natural language processing?
NLP is a subset of machine learning (ML) that understands, interprets, and generates human language in text or voice. Various applications of this technology include machine translation, automatic summarization, information extraction, and text and voice processing. Some of the most famous examples include spell checkers, Google’s predictive suggestions, Siri, Alexa, and chatbots. But we’re interested in how NLP interacts with contract creation and management.
The evolution of NLP in contract management
NLP initially improved the categorization of documents and search features in CLM solutions and document management platforms. Faster and more accurate searches are significant. However, this feature is only the beginning of what NLP and other ML models can do.
A more advanced use of NLP is reading and understanding clauses and subclauses in contracts. Currently, most automated contract solutions on the market offer a relatively shallow approach to this use case. For example, ML models can identify and alert lawyers or business professionals to ambiguous language in their contracts. From there, lawyers negotiate changes to the contract language to avoid ambiguity and mitigate the risk of future conflict. Unfortunately, this use case, in this form, has limited value for veteran lawyers handling high-value agreements.
This use case is not artificial intelligence (AI) either. These solutions must offer the contract drafters alternative language or make automatic changes to be considered AI. Simply highlighting potentially problematic language is beneficial but falls short.
Another example of NLP in contracting is contract creation. For example, NLP allows solutions to create drafts of contracts by pulling information from another source, such as a chatbot, spreadsheet, or form. For formulaic and low-risk contracts, the solution might complete an enforceable agreement. However, higher-value and bespoke contracts require a lawyer’s eyes and insight.
The benefits of NLP in contract management systems
These initial applications of NLP within the contract lifecycle are helpful, particularly regarding high-volume routine contracts. Lawyers and business people spend far too much time wrangling manual contracting processes. Meanwhile, ML models can automate various aspects of routine contracting. The result is improved efficiency, faster turnaround times, more consistent language, and reduced risk of conflicts arising from ambiguous language.
An added benefit is higher employee productivity and morale. High-volume routine contracts consistently pull lawyers and business associates away from their core work. When contract automation handles more manual tasks, they have more time and energy to focus on what matters.
Why contracts are hard to read
There are valid reasons why many use cases of NLP within contract management remain narrow or shallow. One reason is the traditional contract language. In other words, it isn’t easy for the algorithm to understand what the text in the document means.
Researchers have finally looked into why it’s hard for people — not just machines — to understand contract language. Cognitive scientists from MIT analyzed thousands of contracts and found several issues that caused readers trouble: jargon, passive voice, non-standard capitalization, and bad writing.
However, the real culprit that makes contracts hard to understand is center-embedded clauses. That means there’s a clause, often a definition, placed in the middle of the sentence. The researchers found this made it harder for a reader to understand what that provision meant and retain that information.
Another issue these researchers didn’t discuss, but many professionals recognize, is the increasing length and complexity of contracts. Even low-value contracts, such as NDAs, have grown in length over the years.
An additional challenge arises when numerous individuals handle high-volume routine contracts for a single organization. The result is differing contract language, particularly for organizations lacking predetermined standardized language or contract templates.
How Ontra is improving NLP within contracts
Ontra’s goal is to go deeper than the current applications of NLP in contract management. We laid the foundation of that goal with our proprietary Anatomy through which we defined and structured the previously unstructured information within an NDA. Instead of merely mapping provisions, Ontra’s solutions gather meaning from the contract language, moving beyond simple rule-based matching.
Most importantly, Ontra has built human-in-the-loop (HITL) technology. Lawyers’ decisions in the platform continually feed new information into our model. We have access to over 500,000 NDAs — a number that grows every day — giving our model a vast amount of information to learn from and improve speed and results.
A third-party, multinational survey highlighting key findings in contract outsourcing and automation.
As the world around us rapidly changes, how is the contract workflow adapting along with it — and what role do contract outsourcing and contract automation play in this new landscape?