[4P3-OS-8] OS-8

[4P3-OS-8-01] Recognizing Rights and Obligations in Contracts

〇Ruka Funaki^1,2, Yusuke Nagata², Kohei Suenaga², Shinsuke Mori² (1. LegalForce, Inc., 2. Kyoto University)

Keywords:contract, legal document, strcturing text, information extraction, document understanding

A contract is an agreement that binds two or more parties to legally enforceable obligations. For the parties entering into a contract, it is critical for them to understand precisely their obligations. However, for many parties, it can be difficult and costly to achieve the understanding they need to avoid legal liability for a breach of their contract, especially when the contract is long and complicated. A solution to this problem is automated information extraction through the use of machine learning. Such a solution would enable parties to understand their obligations more easily and clearly. To this end, we suggest a methodology for information extraction from contracts involving three steps: (1) extraction of spans of parties, rights, obligations, conditions, and exceptions, (2) association of parties with rights and obligations, and (3) association of conditions and exceptions with rights and obligations. To demonstrate the feasibility of our methodology, we have trained machine learning algorithms and have analyzed their efficacy. As we aim to demonstrate, such algorithms likely can indeed help parties better understand their contractual obligations.