Computable contracts for insurance: establishing an insurance-specific controlled natural language - InsurLE
File(s)
Author(s)
Kowalski, Robert
Cummins, John
Davila, Jacinto
Ovenden, David
Type
Journal Article
Abstract
For the most part, insurance contracts are monolithic, text-based documents, written in unstructured natural language interspersed with technical terms. This means that the many provisions embedded in the contract, both explicit and implicit, are difficult for both humans and computers to understand and process. Efforts to make insurance contracts computable have typically resulted in codified representations of contracts that are separate from the contract wording. This creates the problem of verifying whether or not the contract and the codified representation of the contract have the same meaning. In this paper we address this problem by proposing the use of a computable, controlled natural language, InsurLE, which exposes the internal logic of a contract, while preserving much of the syntactic form of the original wording.
InsurLE is a domain-specific extension of LE (Logical English), which in turn is ‘syntactic sugar’ for the purely logical subset of the computer language Prolog and other logic programming languages. InsurLE extends LE by incorporating an insurance-specific ontology, restrictive relative clauses and a representation of rules and exceptions that mirrors their representation in existing insurance contract wordings. However, like LE and unlike many existing contract wordings, InsurLE avoids complex syntactic structures, which can make insurance contracts difficult to understand and process. As a consequence, contracts written in InsurLE can not only be executed by computer, but can also be easier for laypeople to understand.
InsurLE is a domain-specific extension of LE (Logical English), which in turn is ‘syntactic sugar’ for the purely logical subset of the computer language Prolog and other logic programming languages. InsurLE extends LE by incorporating an insurance-specific ontology, restrictive relative clauses and a representation of rules and exceptions that mirrors their representation in existing insurance contract wordings. However, like LE and unlike many existing contract wordings, InsurLE avoids complex syntactic structures, which can make insurance contracts difficult to understand and process. As a consequence, contracts written in InsurLE can not only be executed by computer, but can also be easier for laypeople to understand.
Date Issued
2024-10-15
Date Acceptance
2024-01-15
Citation
British Insurance Law Association Journal, 2024, 2024 (136)
Publisher
BILA
Journal / Book Title
British Insurance Law Association Journal
Volume
2024
Issue
136
Copyright Statement
© 2024 The Author(s). Published by BILA. This accepted manuscript is available under a Creative Commons Attribution NonCommercial ShareAlike (CC BY-NC-SA)
Identifier
10.1177/20555636211072560
Subjects
computable contract
controlled natural language
Prolog
logic programming
commercial insurance
Publication Status
Published
Date Publish Online
2024-10-14