‘I agree,’ says the button. There’s another button that says ‘Cancel’. You start reading the end-user agreement. By the end of the first clause, you’re thinking of your evening plans. Anyway, you need this software. Click. You’ve agreed. 

Software, mortgage, internet service, rent: our actions are regulated by contracts. Not only are they a challenge for the layperson staring at the dotted line or hovering over the ‘agree’ button, they’re hard work for the lawyers who draft and analyse them. No contract exists in isolation: They link up to other agreements and legislation. Ultimately, they affect people. 

Can computer systems help? 

Until recently, lawyers’ use of computers was largely limited to word processing. This is changing fast, with a recent burgeoning of specialised software for the legal profession. Nevertheless, there are huge challenges that need to be addressed before lawyers can enjoy the level of software support that, say, architects have come to expect from a CAD package. Part of the problem is that contracts, unlike buildings, do not involve geometry, materials engineering and principles of physics; they involve language and language is ambiguous. 

A good engineering software package should be able to tell you that the bridge you’re designing could collapse under a certain weight. The equivalent legal conundrum would be detecting an inconsistency in a contract, say, a case where one clause stipulates something that another clause, or a law, explicitly rules out. Flagging that up requires your software to “understand” the language in which the document is written.

Once software gets past the language barrier, the next obstacle is domain knowledge. Suppose you lease an apartment. There’s a clause in your contract prohibiting you from subletting to someone else without the owner’s consent. But another clause gives the owner the right to refuse entry to the building to anyone of their choice.

There’s a potential conflict there: Subletting to someone implies giving them automatic access to the building. The University of Malta and Loqus Ltd, through a collaboration made possible by the University Research and Innovation Development Trust, have been working together building an intelligent contract drafting tool to support lawyers and notaries. This experimental tool is integrated with standard word processing software. To a lesser extent, the tool also targets the lay person trying to make head or tail of a contract they might be about to sign (or click). We use natural language processing and AI techniques to achieve a partial interpretation automatically. 

Our vision is to incorporate additional functionality such as a law-sensitive ‘track changes’ feature and comparison among versions of a document. For the perplexed, hovering with their mouse over the ‘agree’ button, future extensions include automatic simplification and explanation. 

Ultimately, our goal is to leverage developments in AI to maximise efficiency for the professional, while facilitating understanding for the non-professional. 

The research presented in this article was funded by Loqus Ltd through the Research, Innovation and Development Trust of the University of Malta. Dr Gatt is from the Institute of Linguistics and Language Technologies and Prof. Pace is from the department of Computer Science at the University of Malta.

Did you know?

• A smart contract is an agreement that automates what usually would be a manual process or task. For example, it can automatically calculate what payments need to be made between two parties and arranges for those payments to be made.

• The self-executing contract is handled by a software program that translates a contract into logic statements of the type “if this, then that”.

• Not all parts of a smart contract need to be automated, but the aim of smart contracts is to allow processes under an agreement to happen more efficiently in an unambiguous and predictable way.

• Smart contracts not only define rules and penalties around an agreement in the same way that a traditional contract does, but also automatically enforces those obligations.

For more trivia see: www.um.edu.mt/think

Sound bites

• A recent study suggests that employees can be better motivated to act more securely when it comes to protecting the information within a business and how employees respond to threats. Security managers and supervisors can have greater success by avoiding authoritative commands, and instead create security messages that are relatable and provide options in terms of possible actions that can be taken.


• Ongoing research is attempting to detect vulnerabilities of devices and networks that are used in smart homes (Internet of Things devices). Off-the-shelf devices, such as baby monitors, home security cameras, doorbells and thermostats, were easily co-opted by cyber researchers. The research aims at making manufacturers aware of the security issues and encouraging them to stop using systems that make the devices vulnerable to attacks.


• For more interesting science news listen to Radio Mocha every Saturday at 11.05am on Radju Malta 93.7FM.


Independent journalism costs money. Support Times of Malta for the price of a coffee.

Support Us